UNIT - I: Introduction to Artificial Intelligence

Syllabus:

Introduction, history, intelligent systems, foundations of AI, applications, tic-tac-toe

game playing, current trends in AI.

Outcomes:

Student will be able to:

  define the concept of Artificial Intelligence

  explain brief history that contributed ideas and techniques to Artificial Intelligence
  outline current trends in AI
 interpret the steps in solving of tic-tac-toe problem

Introduction:
  The field of artificial intelligence, or AI, attempts to understand intelligent entities.

Computers with human-level intelligence (or better) would have a huge impact on our everyday
lives and on the future course of civilization.
  
Computers have unlimited potential for intelligence.

AI currently encompasses a huge variety of subfields, from general-purpose areas such as
perception and logical reasoning, to specific tasks such as playingchess, proving
mathematical theorems, writing poetry, and diagnosing diseases.

Definitions of artificial intelligence:

  Definitions of AI vary along following dimensions.
  thought processes and reasoning
  address behaviour
  measure success in terms of human performance,
 measure against an ideal concept of intelligence, which we will call rationality
  A system is “rational” if it does the right thing.
 
 Definitions are organized into four categories:






  humans: The exciting new effort to make computers think
Systems that think like
machines with minds.

  The study of the computations that make it possible to
Systems that think rationally:
perceive, reason, and act.

Systems that act like humans: The study of how to make computers do things at which, at
the moment, people are better.
 
 of computer science that is concerned with the
Systems that act rationally: The branch
automation of intelligent behaviour.


 The Turing Test, proposed by Alan Turing (1950), was designed to provide a satisfactory
operational definition of intelligence.

Turing defined intelligent behaviour as the ability to achieve human-level performance in all
 cognitive tasks, sufficient to fool an interrogator.

The test he proposed is that the computer should be interrogated by a human via a
teletype, and passes thetest if the interrogator cannot tell if there is a computer or a
human at the other end.
  The computer would need to possess the following capabilities:
 natural language processing to enable it to communicate successfully in English
 (or some other human language);
 knowledge representation to store information provided before or during the
 interrogation;
 automated reasoning to use the stored information to answer questions and to
 draw new conclusions
 machine learning to adapt to new circumstances and to detect and extrapolate
 patterns.

 and the
Turing's test deliberately avoided direct physical interaction between the interrogator
computer, because physical simulation of a person is unnecessary for intelligence.


 
To say that a given program thinks like a human, we must have some way of determining
how humans think. We need to get inside the actual workings of human minds.
  There are two ways for this:
  through introspection—trying to catch our own thoughts as they go by.
  through psychological experiments

Once we have a sufficiently precise theory of the mind, it becomes possible to express the
theory as a computer program. If the program's input/output and timing behaviour
matches human behaviour, that is evidence that some of the program's mechanisms may
also be operating in humans.


 The Greek philosopher Aristotle was  one of the first to attempt to codify "right thinking," that
is, irrefutable reasoning processes.

His famous syllogisms provided patterns for argument structures that always gave correct
 
conclusions given correct premises. For example, "Socrates is a man; all men are mortal;
therefore Socrates is mortal."

 These laws of thought were supposed to govern the operation of the mind, and initiated the
field of logic.
  There are two main obstacles to this approach.
 First, it is not easy to take informal knowledge and state it in the formal terms
required by logical notation, particularly when the knowledge is less than 100%
certain.
  Second, there is a big difference between being able to solve a problem "in
principle" and doing so in practice.
Acting rationally: The rational agent approach:
  Acting rationally means acting so as to achieve one's goals, given one's beliefs. 
  An agent is just something that perceives and acts.



  In the ‘laws of
In this approach, AI is viewed as the study and construction of rational agents.
thought" approach to AI, the whole emphasis was on correct inferences.

Making correct inferences is sometimes part of being a rational agent, because one way to
act rationally is to reason logically to the
conclusion that a given action will achieve one's
 goals, and then to act on that conclusion.

 inference is only a useful mechanism for achieving rationality, and not a necessary
Correct
one.
Foundations of Artificial Intelligence:

AI itself is ayoung field; it has inherited many ideas, viewpoints, and techniques from other
 disciplines.

Over 2000 years of tradition in philosophy, theories of reasoning and learning have
emerged, along  with the viewpoint that the mind is constituted by the operation of a
 physical system.

 
From over 400 years of mathematics, we have formal theories of logic, probability, decision
making, and computation.

 From psychology, we have the tools with which to investigate the human mind, and a
scientific language within which to express the resulting theories.
  From linguistics, we have theories of the structure and meaning of language.
 
From computer science, we have the tools with which to make AI a reality.
Philosophy (428 B.C.-present)

  of (at least part of) the mind,
We have the idea of a set of rules that can describe the working
the next step is to consider the mind as a physical system.

 Mental processes and consciousness are therefore part of the physical world, but inherently
unknowable; they are beyond rational understanding.
  Some philosophers critical of AI have adopted exactly this position.

Barring these possible objections to the aims of AI, philosophy had thus established a
tradition in which the mind was conceived of asa physical device operating principally by
 reasoning with the knowledge that it contained.

 All knowledge can be characterized by logical theories connected, ultimately, to observation
sentences that correspond to sensory inputs.

Confirmation theory of Rudolf Carnap and Carl Hempel attempted to establish the nature
 more general theories—
of the connection between the observation sentences and the
understand how knowledge can be acquired from experience.
Mathematics (c. 800-present)
  AI used mathematical formalization in three main areas: computation, logic, and probability. 
 
First-order logic is used today as the most basic knowledge representation system.
  Theory of reference that shows how to relate the objects in a logic to objects in the real world.

There are some functions on the integers that cannot be represented by an algorithm— that is,
 they cannot be computed.

 This motivated Alan Turing (1912-1954) to try to characterize exactly which functions are
capable of being computed.

Church-Turing thesis, which states that the Turing machine (Turing, 1936) is capable of
computing any computable function, is generally accepted as providing a sufficient
definition.
Psychology (1879-present)

 The view that the brainpossesses and processes information, is the principal characteristic of
cognitive psychology,
  The three key steps of a knowledge-based agent:

  The stimulus must be translated into an internal representation,

 The representation is manipulated by cognitive processes to derive new internal
 representations,
 These are in turn retranslated back into action.

  For artificial intelligence to succeed, we need two things: intelligence and an artifact.

 
The computer has been unanimously acclaimed as the artifact with the best chance of
demonstrating intelligence.

 AI also owes a debt to the software side of computer science, which has supplied the 
operating systems, programming languages, and tools needed to write modern programs.

AI has pioneered many ideas that have made their way back to "mainstream" computer
science, including time sharing, interactive interpreters, the linked list data type,
automatic storage management, and some of the key concepts of object-oriented
programming  and integrated program development environments with graphical user
interfaces.
Linguistics (1957-present)

 Understanding language requires an understanding of the subject matter and context, not just
an understanding of the structure of sentences.

Much of the early work in knowledge representation (the study of how to put knowledge
into a form that a computer can reason with) was tied to language and informed by

research in linguistics, which was  connected in turn to decades of work on the
philosophical analysis of language.

Modern linguistics and AI were "born" at about the same time, so linguistics does not
play a large foundational role in the growth of AI. Instead, the two grew up together,
intersecting in a hybrid field called computational linguistics or natural language
processing, which concentrates on the problem of language use.
The History of Artificial Intelligence
The gestation of artificial intelligence (1943-1956)

AI was done by Warren McCulloch and
The first work that is now generally recognized as
Walter Pitts (1943). They drew on three sources:
  knowledge of the basic physiology
  function of neurons in the brain
  the formal analysis of propositional logic

Turing's theory of computation: They proposed a model of artificial neurons in which
each neuron is characterized as being "on" or "off," with a switch to  "on" occurring in
 response to stimulation by a sufficient number of neighbouring neurons.

 Claude Shannon (1950) and Alan Turing(1953) were writing chess programs for von
Neumann-style conventional computers.

 Marvin
At the same time, two graduate students in the Princeton mathematics department,
Minsky and Dean Edmonds, built the first neural network computer in 1951.
Early enthusiasm, great expectations (1952-1969)

Newell and Simon's early success was followed up with the General Problem Solver, or
 GPS. Unlike Logic Theorist,this program was designed from the start to imitate human
problem-solving protocols.

Within the limited class of puzzles it could handle, it turned out that the order in which
the program considered subgoals and possible actions was similar to the way humans

approached the same problems. Thus, GPS was probably the first program to embody the
"thinking humanly" approach.

  Prover. Like the Logic Theorist,
Herbert Gelernter (1959) constructed the Geometry Theorem
it proved theorems using explicitly represented axioms.

  too many possible reasoning paths to follow, most of
Gelernter soon found that there were
which turned out to be dead ends.

Arthur Samuel wrote a series of programs for checkers (draughts) that eventually learned
to play tournament-level checkers. Along the way, he disproved the idea that computers

can only do what  they are told to, as his program quickly learned to play a better game
than its creator.

 
McCarthy defined the high-level language Lisp, which was to become the dominant AI
programming language.

Rosenblatt proved the famous perceptron convergence theorem, showing that his learning 
algorithm could adjust the connection strengths of a perceptron to match any input data.
A dose of reality (1966-1974)

Most of the early AI programs worked by representing the basic facts about a problem
 and trying out a series ofsteps to solve it, combining different combinations of steps until
the right one was found.
  The early programs were feasible only because micro worlds contained very few objects.

 Before the theory of NP-completeness was developed, it was widely thought that "scaling up"
to larger problems was simply a matter of faster hardware and larger memories.

 Resolution theorem proving, was soon  dampened when researchers failed to prove theorems
involving more than a few dozen facts.
 
 A two-input perceptron could not be trained to recognize when its two inputs were different.

Althoughtheir results did not apply to more complex, multilayer networks, solved the
problem.
 
The new back-propagation learning algorithms for multilayer networks that were to cause an
enormous resurgence in neural net research.


General-purpose search mechanism performs elementary reasoning steps to find complete
 
solutions. Such approaches are called weak methods, because they use weak information
about the domain.

The only way around this is to use knowledge more suited to making larger reasoning steps
 and to solving typically occurring cases in narrow areas of expertise.

 The DENDRAL program solves the problem of inferring molecular structure from the
information provided by a mass spectrometer.

The input to the program consists of the elementary formula of the molecule, and the
 mass spectrum giving the masses of the  various fragments of the molecule generated
when it is bombarded by an electron beam.
  The naive version of the program generated all possible structures consistent with the
  
formula, and then predicted what mass spectrum would be observed for each.
  
The significance of DENDRAL was that it is the first successful knowledge-intensive
  
System.

 MYCIN diagnoses blood infections. With about 450 rules, MYCIN  was able to perform as
well as some experts, and considerably better than junior doctors.
  It also contained two major differences from DENDRAL.:
 First, unlike the DENDRAL rules, no general theoretical model existed from
which the MYCIN rules could be deduced. They had to be acquired from
extensive interviewing of experts, who in turn acquired them from direct
 experience of cases.
 Second, the rules had to reflect the uncertainty associated with medical
knowledge. MYCIN incorporated a calculus of uncertainty called certainty factors.


In 1981, the Japanese announced the "Fifth Generation" project, a 10-year plan to build
 intelligent 
computers running Prolog in much the same way that ordinary computers run
machine code.
  It has the ability to make millions of inferences per second.

  to achieve full-scale natural language understanding, among other
The project proposed
ambitious goals.

The booming AI industry also included companies such as Carnegie Group, Inference,
Intellicorp, and Teknowledge that offered the software tools to build expert systems, and
hardware companies such as Lisp Machines Inc., Texas Instruments, Symbolics, and 
Xerox that were building workstations optimized for the development of Lisp programs.
The return of neural networks (1986-present)

 Back-propagation learning was applied to many learning problems in computer science and
psychology, and the widespread dissemination of the results in Parallel Distributed
Processing.
 
In recent years, approaches were based on hidden Markov models (HMMs) which are
based on a rigorous mathematical theory. These are generated by a process of training on
a large corpus of real speech data.
Intelligent Systems

Intelligence:The ability of a system to calculate, reason, perceive relationships and
analogies, learn from experience, store and retrieve information from memory, solve
 
problems, comprehend complex ideas, use natural language fluently, classify, generalize,
and adapt new situations.

 Intelligent systems
are technologically advanced machines that perceive and respond to the
world around them.

 An intelligent system is a computer-based system that can represent reason about, and
interpret data.
An intelligent system is a system with artificial intelligence.


 Reasoning − It is the set of
processes that enables us to provide basis for judgement, making
decisions, and prediction.

Learning − It is the activity of gaining
 knowledge or skill by studying, practising, being
 taught, or experiencing something.

Problem solving- It is the process of selecting the  best suitable alternative out of multiple
 alternatives to reach the desired goal are available.

Perception − It is the process of acquiring, interpreting, selecting, and organizing sensory
information. Perception presumes sensing. In humans, perception is aided by sensory
 
organs. In the domain of AI, perception mechanism puts the data acquired by the sensors
together in a meaningful manner.

Linguistic Intelligence − It is one’s ability to use, comprehend, speak,
and write the verbal
and written language. It is important in interpersonal communication.
Applications of AI
Air Operations Division (AOD) - uses AI for the rule based expert systems.
The AOD has use for artificial intelligence for surrogate operators for combat and
training simulators, mission management aids, support systems for tactical
decision making, and post processing of the simulator data into symbolic
summaries.
  Airplane simulators are using artificial intelligence in order to process the data
 taken from simulated flights.
 Simulated aircraft warfare computers are able to come up with the best success
 scenarios in these situations

  have created many tools to solve the most difficult
Computer Science: AI researchers
problems in computer science.
time sharing ,
interactive interpreters,
graphical user interfaces and the computer mouse,
rapid development environments,
the linked list data structure,
automatic storage management,
symbolic programming,
functional programming,
dynamic programming
object-oriented programming

Education: Number of companies that create robots to teach subjects to children ranging
 from biology to computer science.
Intelligent tutoring systems: An ITS called SHERLOCK teaches Air Force
technicians to diagnose electrical systems problems in aircraft. Another example
is DARPA, Defense Advanced Research Projects Agency, which used AI to
develop a digital tutor to train its Navy recruits in technical skills in a shorter
amount of time.
Finance: Algorithmic Trading involves the use of complex AI systems to make trading
decisions at speeds several orders of magnitudes greater than any human is capable of,
often making millions of trades in a day without any human intervention. Such trading is
called High-frequency Trading. Many banks, funds, and proprietary trading firms now
have entire portfolios which are managed purely by AI systems.
Hospitals and medicine: used as clinical decision support systems for medical diagnosis.
Computer-aided interpretation of medical images help scan digital images, e.g.
from computed tomography, for typical appearances and to highlight conspicuous
sections, such as possible diseases. A typical application is the detection of a
tumor.
Heart sound analysis
Companion robots for the care of the elderly
  Mining medical records to provide more useful information.
  Design treatment plans.
  Assist in repetitive jobs including medication management.
  Provide consultations.
  Drug creation[27]
  Using avatars in place of patients for clinical training[28]
  Predict the likelihood of death from surgical procedures
  Predict HIV progression
  
Media and E-commerce: Typical use case scenarios include:
analysis of images using object recognition or face recognition techniques,
analysis of video for recognizing relevant scenes, objects or faces.
  facilitation of media search,
  creation of a set of descriptive keywords for a media item,
 media content policy monitoring (such as verifying the suitability of content for a
 particular TV viewing time),
speech to text for archival or other purposes, and the detection of logos,
  products or celebrity faces for the placement of relevant advertisements
 AI is also widely used in E-commerce Industry for applications like Visual
search, Visually similar recommendation, Chatbots, Automated product
tagging etc

Music: At Sony CSL Research Laboratory, their Flow Machines software has created pop
 songs by learning music styles from a huge database of songs. By analyzing  unique
combinations of styles and optimizing techniques, it can compose in any style.

News, publishing and writing: Artificial intelligence is used to turn structured data into
intelligent comments and recommendations in natural language. We can be able to write
financial reports, executive summaries, personalized sales or marketing documents and
more at a speed of thousands of pages per second and in multiple languages including
English, Spanish, French & German.

Online and telephone customer service: Artificial intelligence is implemented in
automated online assistants which uses natural language processing.
Sensors: Artificial Intelligence has been combined with many sensor technologies, such
as Digital Spectrometry TM which enables many applications such as at home water
quality monitoring.
Telecommunications maintenance: Many telecommunications companies make use of
heuristic search in the management of their workforces, for example BT Group has
deployed heuristic search in a scheduling application that provides the work schedules of
20,000 engineers.
Toys and games: AI has been applied to video games, for example video game bots, which
are designed to stand in as opponents where humans aren't available or desired.
Transportation: Fuzzy Logic controllers have been developed for automatic gearboxes
in automobiles. For example, the 2006 Audi TT, feature the DSP transmission which
utilizes Fuzzy Logic.

AI trends in various sectors

  Healthcare:
 AI and ML technology has been particularly useful in the healthcare industry
because it generates massive amounts of data to train with and enables algorithms
 to spot patterns faster than human analysts.
Medecision developed an algorithm that detects 8 variables in diabetes patients to
determine if hospitalization is required.
 An app called BiliScreen utilizes a smartphone camera, ML tools, and computer
vision algorithms to detect increased levels of bilirubin in the sclera (white
portion) of a person’s eye, which is used to screen people for pancreatic cancer.
This cancer has no telltale symptoms, hence it has one of the worst prognoses of
all cancers.
NuMedii, a biopharma company, has developed a platform called Artificial
Intelligence for Drug Discovery (AIDD), which uses big data and AI to detect the
link between diseases and drugs at the systems level.
GNS Healthcare uses ML algorithms to match patients with the most effective
treatments for them.
  Entertainment: 

  A familiar application of AI in everyday life is seen with services like Netflix or
Amazon, wherein ML algorithms analyze the user’s activity and compare it with
that of other users to determine which shows or products to recommend. The
algorithms are becoming intelligent with time—to the extent of understanding that
a user may want to buy a product as a gift and not for him/her, or that different
family members have different watching preferences.
  
Finance:
Financial services companies use AI-based natural language processing tools to
analyze brand sentiment from social media platforms and provide actionable
advice.
Investment companies like Aidya and Nomura Securities use AI algorithms to
conduct trading autonomously and robo-traders to conduct high-frequency trading
for greater profits, respectively.
Fintech firms like Kensho and ForwardLane use AI-powered B2C robo-advisors
to augment rebalancing decisions and portfolio management performed by human
analysts. Wealthfront uses AI algorithms to track account activity and help
financial advisors customize their advice.
 Chatbots, powered by natural language processing, can serve banking customers
quickly and efficiently by answering common queries and providing information
promptly.

Fraud detection is an important application of AI in financial services. For
example, Mastercard uses Decision Intelligence technology to analyze various
data points to detect fraudulent transactions, improve real-time approval accuracy,
and reduce false declines.
  Data security: 

Cyber attacks are becoming a growing reality with the move to a digital world.
There are also concerns about AI programs themselves turning against systems.
Automatic exploit generation (AEG) is a bot that can determine whether a
software bug, which may cause security issues, is exploitable. If a vulnerability is
found, the bot automatically secures it. AEG systems help develop automated
 signature generation algorithms that can predict the likelihood of cyberattacks.



 PatternEx and MIT’s Computer Science and Artificial Intelligence Laboratory
(CSAIL) have developed an AI platform called AI2 which claims to predict cyber
attacks better than existing systems. The platform uses Active Contextual
Modeling, a continuous feedback loop between a human analyst and the AI
system, to provide an attack detection rate that is better than ML-only solutions by
a factor of 10.
Deep Instinct, an institutional intelligence company, says that malware code varies
between 2%-10% in every iteration and that its AI model is able to handle the
variations and accurately predict which files are malware.
  Manufacturing: 

Landing AI claims to have created machine-vision tools to find microscopic

defects in objects like circuit boards using an ML algorithm trained using tiny
volumes of sample images. In the future, self-driving robots may be created which
can move finished goods around without endangering anyone or anything around.
Robots in factories are often stationary but are still in danger of crashing into
objects around it. A new concept called collaborative robots or “cobots, enabled
by AI, can take instructions from humans, including instructions that the robot has
not been previously exposed to, and work productively with them.
 AI algorithms can influence the manufacturing supply chain by detecting the
patterns of demand for products across geographies, socioeconomic segments, and
time, and predicting market demand. This, in turn, will affect inventory, raw
material sourcing, financing decisions, human staffing, energy consumption, and
maintenance of equipment.

 AI tools help in predicting malfunctions and breakdown of equipment and taking
or recommending preemptive actions as well as tracking operating conditions and
performance of factory tooling.
  Automotive industry 

Tesla introduced TeslaBot, an intelligent virtual assistant integrated with Tesla

models S and X, allows users to interact with their car from their phone or
desktop.
Uber AI Labs is working on developing self-driven cars with the help of the best
engineers and scientists. Uber has already tested a batch of self-driving cars in
2016.
Nvidia has partnered with Volkswagen to develop “intelligent co-pilot systems” in
cars that will enable safety warnings, gesture control, and voice and facial
recognition.
Ericsson predicts that 5G technology will improve vehicle-to-vehicle
communication wherein sensors will be implanted in airport runways, railways,
and roads.
Tic Tac Toe Game playing

The game Tic Tac Toe is also known as Noughts and Crosses or Xs and Os ,the player
needs to take turns marking the spaces in a 3x3 grid with their own marks, if 3
 these moves get won.
consecutive marks (Horizontal, Vertical, Diagonal) are formed then the player who owns

Assume ,
Player 1 - X
Player 2 – O
 
Board's Data Structure:










 
The cells could be represent as Center square, Corner,Edge as like below:
 Assignment-Cum-Tutorial Questions

I) Objective Questions

1. Define Artificial Intelligence____________________________________

2. A system is said to be _____________ if it does the right thing, given known facts.
3. The study of how to make computers do things at which, at the moment, people are better-is
a characteristic of_____. [ ]
(a) Systems that think like humans (b) Systems that think rationally
(c) Systems that act like humans (d) Systems that act rationally
4. The “Turing Test approach” is used to test whether a system ___________ [ ]
(a) think like humans (b) think rationally (c) act like humans (d) act rationally
5. The “Cognitive Modelling” approach is used to test whether a system ____. [ ]
(a) think like humans (b) think rationally (c) act like humans (d) act rationally
6. The “Laws of Thought” approach is used to test whether a system ____. [ ]
(a) think like humans (b) think rationally (c) act like humans (d) act rationally
7. For artificial intelligence to succeed, we need_____. [ ]
(a) intelligence (b) artifact (c) both a & b (d) none
8. Understanding language requires an understanding of _______. [ ]
(a) subject matter (b) context (c) structure of sentences (d) only a & b
9. GPS was probably the first program to embody _________ approach. [ ]
(a) thinking humanly (b) acting humanly (c) thinking rationally (d) acting rationally
10. The __________program solves the problem of inferring molecular structure from the
information provided by a mass spectrometer. [ ]
(a) MYCIN (b) DENDRAL (c) both a & b (d) none
11. __________ diagnoses blood infections. [ ]
(a) MYCIN (b) DENDRAL (c) both a & b (d) none
12. ____________teaches Air Force technicians to diagnose electrical systems problems in
aircraft. [ ]
(a) SHERLOCK (b) DARPA (c) DENDRAL (d) none

13. DARPA stands for_____________.

II) Descriptive Questions

1. Summarize AI definition categories?

2. Illustrate the capabilities that a computer must possess to pass Turing Test?

3. Explain the areas from which Artificial Intelligence laid its foundation?

4. Explain the history of Artificial Intelligence?

5. List the applications of Artificial Intelligence?

6. Outline the current trends in Artificial Intelligence?

7. What is an Intelligent System? Explain its characteristics?

8. Interpret the steps to solve tic-tac-toe problem.

 Artificial Intelligence Unit-II
UNIT - II: Problem solving and Game playing
Syllabus:

Problem solving: state-space search and control strategies: Introduction,

general problem solving, characteristics of problem, exhaustive searches,
heuristic search techniques, iterative-deepening a*, problem reduction,
constraint satisfaction.
Game playing: Introduction, game playing, alpha-beta pruning, two-player
perfect information games.
Outcomes:
Student will be able to:
  Choose the heuristic to solve the problem.
  Classify problems based the characteristics of a problem.
  Illustrate the method to solve a particular problem.
  Provide formal descriptions of various problems.
 Examine alpha-beta cut-offs in Game playing.

General Problem Solving


 To builda system to solve a particular problem, we need to do four
things:
1. Define the problem precisely. This definition must include precise
specifications of what the initial situation(s) will be as well as what
final situations constitute acceptable solutions to the problem.
2 Analyse the problem. A few very important features can have an
immense impact on the appropriateness of various possible
technique(s)for solving the problem.
3. Isolate and represent the task knowledge that is necessary to
solve the problem.
4. Choose the best problem-solving technique(s) and apply it (them)
to the particular problem.
State-Space Search

To build a program that could "Play chess," we would first have to
specify the starting position of the chess hoard, the rules that define
the legal moves,  and the board positions that represent a win for one
 side or the other.

The starting position can he described as an 8-by-8 array where each
 
position contains a symbol standing for the appropriate piece in the
official chess opening position.

 We can define as our goal any board position in which the opponent does
not have a legal move and his or her king is under attack.

The legal moves provide the way of getting from the initial state to a goal
state. They can he described easily as a set of rules consisting of two
 Artificial Intelligence Unit-II
parts: a left side that serves as a pattern to he matched against the
current board position and a right side that describes the change to be
made to the board position in reflect the move.


We have to write a very large number of rules since there have to be a
separate rule for each of the 10120 possible board positions. Using so
 many rules poses two serious practical difficulties:
 No person could ever supply a complete set of such rules. It would
 take too long and could certainly not be done without mistakes.
 No program could easily handle all those rules. Although
hashing scheme could be used to find the relevant rules for
each move fairly quickly, just storing that many rules poses
 serious difficulties.

We should write the rules describing the legal moves in as general a
way possible. The convenient  notation for describing patterns and
substitutions is as follows:


A problem can be defined in a “State Space”,  where each state
 corresponds to a legal position of the board.
 We can start at an initial state using a set of rules to move from
one state to another, and attempting to end up in one of a set of
 final states.

The state space representation forms the basis of most of the Al
methods. Its structure 
corresponds to the structure of problem solving
in two important ways:
 Artificial Intelligence Unit-II

 It allows formal definition of a problem as the need to convert

some given situation into some desired situation using a set of
 permissible operations.
 It permits us to define the process of solving a particular
problem as a combination of known techniques) each
represented as a role defining a single step in the space) and
search, the general technique of exploring the space to try to
find some path from the current state to a goal state. Search is
a very important process in the solution of hard problems for
which no more direct techniques are available.
Water Jug Problem

Consider the following problem: A Water Jug Problem: You are given
two jugs, a 4-gallon one and a 3-gallon one, a pump which has
unlimited water which you can use to fill the jug, and the ground on

which water may be poured. Neither jug has any measuring markings 
on it. How can you get exactly 2 gallons of water in the 4-gallon jug?

State Representation and Initial State – we will represent a state of
the problem as a tuple (x, y) where x represents the amount of water
in the 4-gallon jug and y represents the  amount of water in the 3-
gallon jug. Note 0 ≤ x ≤ 4, and 0 ≤ y ≤ 3.
  Our Initial state: (0,0) Goal state = (2,y) where 0 ≤ y ≤ 3.

  to make some assumptions not mentioned in the
To solve this we have
problem. They are:
1. We can fill a jug from the pump.
2. We can pour water out of a jug to the ground.
3. We can pour water from one jug to another.
4. There is no measuring device available. To solve the water jug
problem, all we need is a control structure that loops through a simple
cycle in which some rule whose left side matches the current state is
chosen. The appropriate change to the state is made as described in the
corresponding right side, and the resulting state is checked to see ii ii
corresponds to a goal state. As long as it does not, the cycle continues.
Clearly the speed with which the problem gets solved depends on the
mechanism that is used to select the next operation to be performed.

We must define a set of operators that will take us from one state to
another:
Artificial Intelligence Unit-II
 Artificial Intelligence Unit-II


The first step toward the design of a program to solve a problem must be
the creation of a formal and manipulable description of the problem itself.

We should be able to write programs that can themselves produce such
formal descriptions from informal ones. This process is called
 Operationalization.

 In order to provide a formal description of a problem, we must do the
following:
 Define a state space that contains all the possible configurations
of the relevant objects (and perhaps some impossible ones). It is, of
course, possible to define this space without explicitly enumerating
 all of the states it contains.
 Specify one or more states within that space that describes
possible situations from which the problem-solving process may
 start. These states are called the Initial states.
 Specify one or more states that would be acceptable as solutions to
 the problem. These states are called goal states.
 Specify a set of rules that describe the actions (operators) available.



The problem can then be solved by using the rules, in combination
with an appropriate Control Strategy, to move through the problem

space until a path from an initial state to a goal state is found. Thus
the process of search is fundamental to the problem-solving process.
Control Strategies

It is important to decide which rule to apply next during the process of
  when more
searching for a solution to a problem. This question arises
than one will have its left side match the current state.

 The decisions made  will have a crucial impact on how quickly a problem
is finally solved.
  The following are the requirements of a control strategy:
 The first requirement of a good control strategy is that it causes
motion. Consider again the water jug problem. Suppose we
implemented the simple control strategy of starting each time at
the top of the list of rules and choosing the first applicable one.
If we did that, we would never solve the problem. We would
continue indefinitely filling the 4-gallon jug with water. Control
 strategies that do not cause motion will never lead to a solution.
 The second requirement of a good control strategy is that it be
systematic. Let us consider a simple control strategy for the water
jug problem. On each cycle, choose at random from among the
applicable rules. This strategy is better than the first. It
 Artificial Intelligence Unit-II
causes motion. It will lead to a solution eventually. Sometimes
we arrive at the same state several times during the process
and to use many more steps than are necessary. Because the
control strategy is not systematic, we may explore a particular
useless sequence of operators several times before we finally
find a solution. The requirement that a control strategy be
systematic corresponds to the need for global motion (over the
course of several steps) as well as for local motion (over the
course of a single step).
Breadth First Search

For each leaf node, generate all its successors by applying all the rules
that are appropriate. Continue this process until some rule produces
a goal state. This process, called Breadth-First search.

 Algorithm: Breadth-First Search
1. Create a variable called NODE-LIST and set it to the initial state.
2. Until a goal state is found or NODE-LIST is empty do:
(a) Remove the first element from NODE-LIST and call it E. If

(b) For each way that each rule can match the state described in
E do:
i. Apply the rule to generate a new state.
ii. If the new state is a goal state, quit and return this state.
iii. Otherwise, add the new state to the end of' NODE-LIST.

  Advantages of Breadth-First Search:

  Breadth-first search will not get trapped exploring a blind alley.
 If there is a solution, then breadth-first search is guaranteed to
find it. Furthermore, if there are multiple solutions, then a
minimal solution will be found. This is guaranteed by the fact
that longer paths are never explored until all shorter ones have
already been examined.
 Artificial Intelligence Unit-II
  Disadvantages of Breadth-First Search:
 All of the tree that has so far been generated must be stored
 which requires more memory.
 In breadth-first search, all parts of the tree must be examined to
level n before any nodes on level n + 1 can be examined. This is
particularly significant if many acceptable solutions exist.
Depth First Search

  it yields a solution or
We could pursue a single branch of the tree until
until a decision to terminate the path is made.

 It makes sense to terminate a path if it reaches a dead-end, produces a
previous state, or becomes longer than some prespecified "futility" limit.

In such a case backtracking occurs. The most recently created state from
which alternative moves are available will be revisited and a new state
will he created. This form of backtracking is called Chronological

Backtracking because the order in which steps arc undone depends  only
on the temporal sequence in which the steps were originally made.

 procedure we have just described is called Depth-First
The search
search.

 Algorithm: Depth-First Search
1. If the initial state is a goal state, quit and return success.
2. Otherwise, do the following until success or failure is signalled:
(a) Generate a successor, E, of the initial state. It there are no
more successors,
signal failure.
(b) Call Depth-First Search with E as the initial state.
(c) If success is returned, signal success. Otherwise continue in
this loop.

  Advantages of Depth-First Search:

 Depth-first search requires less memory since only the nodes on
the current path are stored.
 Artificial Intelligence Unit-II

 By chance (or if care is taken in ordering the alternative

successor states), depth first search may find a solution without
examining much of the search space at all. Depth-first search
can stop when one of solution is found.
  Disadvantages of Depth-First Search: 

 Depth- first searching may follow a single unfruitful path for a

 very long time.
 Depth-first search, may find a long path to a solution in one
part of the tree, when a shorter path exists in some other
unexplored part of the tree.
Problem Characteristics

A heuristic is a technique that improves the efficiency of a search
process, possibly by sacrificing claims such as completeness.
Heuristics are like tour guides. They are good to the extent that they
point in generally interesting directions: they are bad to the extent
that they may miss points of interest to particular individuals. But, 
on
 the average, they improve the quality of the paths that are explored.

Heuristic search is a very general method applicable to a large class of
problems; it encompasses a variety of specific techniques, each of
which is particularly effective for a small class of problems. In order to
choose the most appropriate method (or combination of methods for a
particular problem, it  is necessary to analyze the problem along
 several key dimensions:
 Is the problem decomposable into a set of (nearly) independent
 smaller or easier sub problems?
 Can solution steps be ignored or at least undone if they prove
 unwise?
  Is the problem's universe predictable?
 Is good solution to the problem obvious without comparison to all
 other possible solutions?
  Is the desired solution a state of the world or a path to a state?
 Is a large amount of knowledge absolutely required to solve the
 problem, or is knowledge important only to constrain the search?
 Can a computer that is simple given the problem return the solution,
or will the solution of the problem require interaction n between
the computer and a person?
1. Is the Problem Decomposab1e?
 
Suppose we want to solve the problem of computing the expression:
 Artificial Intelligence Unit-II

We can solve this problem by breaking it down into three smaller
 problems each of which we can then solve by using a small collection
of specific rules.

The figure shows the problem tree that will be generated by the
 
process of problem decomposition as it can be exploited by a simple
recursive integration program that works as follows:

 At each step. it checks to see whether the problem it is working on is
immediately solvable. If so, then the answer is returned directly.

If the problem is not easily solvable, the integrator checks to see
whether it can decompose the problem into smaller problems. If it can,
it creates those problems and calls itself recursively on them. Using
this technique ofproblem decomposition we can often solve very large
problems easily.


Example-Blocks  World Problem: Assume that the following operators
are available-


Applying the technique of problem decomposition to this simple blocks 
world example would lead to a solution tree such as that shown below:
 Artificial Intelligence Unit-II


In the figure, goals are underlined. States that have been achieved are
 not underlined. The idea of this solution is to reduce the problem of
getting B on C and A on B to two separate problems.

The first of these new problems, getting B on C. is simple, given the
start state. Simply put B on C. The second sub goal is not quite so
simple. Since the only operators we have allow us to pick up single
  C before we can
blocks at a time, we have to clear off B by removing
pick up A and put it on B This can easily be done.

However, if we now try to combine the two sub solutions into one
  we do first, we will not
solution, we will fail. Regardless of which one
be able to do the second as we had planned.

In this problem, the two sub problems are not independent. They 
interact and those interactions must be considered in order to arrive at
a solution for the entire problem.
2. Can Solution Steps Be Ignored or Undone?

 Theorem Proving: We proceed by first proving lemma that we think  will
be useful. Eventually, we realize that the lemma is no help at all.

  true. Any rules that
Everything we need to prove the theorem is still
could have been applied can still be applied.

  All we have lost
We can just proceed as we should have in the first place.
is the effort that was spent exploring the blind alley.

The 8-Puzzle: The 8-puzzle is a square tray in which are placed eight
square tiles. The remaining ninth square is uncovered. Each tile has a
number on it. A tile that is adjacent to the blank space can he slid into
that space. A game consists of a starting position and a specified goal
position. The goal is to transformthe starting position into a goal
position by sliding the tiles around.
 Artificial Intelligence Unit-II

In attempting to solve the 8-puzzle, we might make a stupid move. We
might start by sliding tile 5 into the empty space. Having done that,

we cannot change our mind and immediately slide tile 6 into  the
empty space since the empty space will essentially have moved.

But we can backtrack and undo the first move, sliding tile 5 back to
where it was. Then we can move tile 6. Mistakes can still be recovered

from but not quite easily.An additional step must be performed to
undo each incorrect step,

The control mechanism for an 8.puzzle solver must keep track of the
 
order in which operations are performed so that the operations can be
undone one at a time if necessary.

Chess: Now consider again the problem of playing chess. Suppose a
chess-playing program makes a stupid move and realizes it a couple
of moves later. It cannot simply play as though it had never made the
stupid move. Nor can it simply back up and start the game over from
that point. All it can do is totry to make the best of the current
 situation and go on from there.

 The three problems—theorem proving, the 8-puzzle, and chess— 
illustrate the differences between three important classes of problems:
• Ignorable(e.g... theorem proving), in which solution steps can be
ignored
• Recoverable(e.g... 8-puzzle), in which solution steps can be undone.
• Irrecoverable(e.g... chess), in which solution steps cannot be
undone

The recoverability of a problem plays an important role in
 
determining the complexity of the control structure necessary for the
problem's solution.
 Ignorable problems can be solved using a simple control structure
that never backtracks. Such a control structure is easy to
 implement.
 Recoverable problems can be solved by a slightly more complicated
control strategy that does sometimes make mistakes. Backtracking
will be necessary to recover from such mistakes, so the control
structure must be implemented using a push-down stack, in which
 decisions are recorded in case they need to be undone later.
 Irrecoverable problems, on the other hand, will need to be solved
by a system that expends a great deal of effort making each
 decision since the decision must be final
3. Is the Universe Predictable?

8-puzzle: Every time we make a move, we know exactly what will
happen. This means that it is possible to plan an entire sequence of
moves and be confident that we know what the resulting state will be.
 Artificial Intelligence Unit-II
We can use planning to avoid having to undo actual moves, although
it still be necessary to backtrack past those moves one at a time
during the planning process. Thus a control structure that allows
backtracking will be necessary.

Play Bridge: One of the decisions we will have to make is which card
to play on the first trick. What we would like to do is to plan the entire
hand before making that first play. But now it is not possible to do
such planning with certainty since we cannot know exactly where all
the cards are what the other players will do on their turns. The best
we can do is to investigate several plans and use probabilities of the
various outcomes to choose a plan that has the highest estimated
 probability of leading to a good score on the hand.

Certain–outcome problems (e.g... 8-puzzle): The open-loop approach
will work fine since the result of an action can he predicted perfectly.
 
Thus, planning can be used to generate a sequence of operators that
is guaranteed to lead to a solution.

Uncertain–outcome problems (e.g... Bridge): Planning can at best
generate a sequence of operators that has a good probability of leading to
a solution. To solve such problems, we need to allow for a process of plan
revision to take place as the plan is carried out and the necessary
feedback is provided. In addition to providing no guarantee of art actual
solution, planning for uncertain -outcome problems has the drawback
that it is often very expensive since the number of solution paths that

need to be explored increases exponentially with the number of points
at which the outcome cannot be predicted.
4. Is a Good Solution Absolute or Relative?

 Simple facts problem: Consider the problem of answering questions
based on a database of simple facts, such as the following:
1. Marcus was a man.
2. Marcus was a Pompeian.
3. Marcus was born in 40 A.D.
4. All men are mortal.
5. All Pompeians died when the volcano erupted in 79 A.D.
6 No mortal lives longer than 150 years.
7.11 is now 1991 A.D.

Suppose we ask the question 'Is Marcus alive?'. By representing each
of these facts in a formal language such as predicate logic, and then
using formal inference methods we can fairly easily derive an answer
to the question. Since all we are interested in is the answer to the
que7sion, it does not matter which path we follow. If we do follow one
path successfully to the answer, there is no reason
 to go back and see
it some other path might also lead to a solution.
 Artificial Intelligence Unit-II


Travelling salesman problem: Our goal is to find the shortest route
that visits each city exactly once. Supposethe cities to be visited and
the distances between them are as shown:


One place the salesman could start is Boston. In that case, one path
that might be followed is shown, which is 8850 miles long. But is this
the solution to the problem? The answer is that we cannot be sure
unless we also try all other paths to make sure that none of them is
shorter. In this case,the first path is definitely not the solution to the
salesman's problem.
 Artificial Intelligence Unit-II


Best-path problems: These are, in general, computationally harder
than Any-path problems.  No heuristic that could possibly miss the
 best solution can be used.

Any-path problems: can often be solved in a reasonable amount of time

by using heuristics that suggest good paths to explore. If the heuristics
are not perfect, the search for a solution may not be as direct as
possible. So a much more exhaustive search will be performed.
5. Is the Solution a State or a Path?

Consistent interpretation for the sentence (NLU): There are several
components of this sentence,each of which, in isolation, may have
 more than one interpretation.
“The Bank president ate a dish of pasta salad with the fork”
  Some of the sources of ambiguity in this sentence are the following:
 The word "bank" may refer either to a financial institution or to
 a side of a river. But only one of these may have a president.
 The word "dish" is the object of the verb "eat." It is possible that
a dish was eaten. But it is more likely that the pasta salad in
 the dish was eaten.
 Pasta salad is a salad containing pasta. But there are other
ways meanings can be formed from pair of nouns. For example,
 dog food does not normally contain dog.

Some search may be required to find a complete interpretation for the
sentence. But to solve the problem of finding the interpretation we
need to produce only the interpretation itself. No record  of the
 processing by which the interpretation was found is necessary.

Water Jug Problem: It is not sufficient to report that we have solved the
problem and that the final state (2, 0). For this kind of problem, what we
found to that
really must report is not the final state but the path that we
state. Thus a statement of a solution to this problem must
 Artificial Intelligence Unit-II
be a sequence of operations called a “plan” that produces the final
state.
  The difference between these problems is: 

  Problems whose solution is a state.

 Problems whose solution is a path to a state.
 
In water jug problem, we must re-describe the states so that each state
represents a partial path to a solution rather than just a single state
of the world.
6. What Is the Role of Knowledge?

Playing chess: Suppose you had unlimited computing power
available. The knowledge that would be required by a perfect program
is very little—just the rules for determining legal moves andsome
 simple control mechanism that implements a search procedure.

Additional knowledge about such things as good strategy, and tactics
 could of course help considerably to constrain the search and speed
up the execution of the program.

Scanning daily newspapers: to decide which are supporting the
Democrats and which are supporting the Republicans in some
upcoming election. Again assuming unlimited computing power, how
 problem? It would have to know such things as:
much knowledge would be required by a computer trying to solve this

• The names of the candidates in each party.

• The fact that if the major thing you want to see done is have taxes
lowered, you are probably supporting the Republicans.
• The fact that it the major thing you want to see done is improved
education for minority students, you are probably supporting the
Democrats.
• The fact that if you are opposed to big government you are probably
supporting the Republicans.
 
These two problems, chess and newspaper story understanding,
illustrate the difference between problems for which a lot of knowledge
is important only to constrain the search for a solution and those for
which a lot of knowledge is required even to be able to recognize a
solution.
7. Does the Task Require Interaction with a Person?

Sometimes it is useful to program computers to solve problems in
ways that the majority of people would not be able to understand.
 
This is fine if the level of the interaction between the computer and its
human users is program-in solution-out.

But increasingly we are building programs that require intermediate
interaction with people, both to provide additional input  to the
program and to provide additional reassurance to the user.
 Artificial Intelligence Unit-II
  We must distinguish between two types of problems:

Solitary: in which the computer is given a problem description and
 
produces an answer with no intermediate communication and with no
demand for an explanation of the reasoning process

Conversational: in which there is intermediate communication 
between a person and the computer, either to provide additional
assistance to the computer or to provide additional information to the
user, or both.


Generic control strategy-Classification: The task here is to examine
an input and then decide which of a set of known classes the input is
an instance of. Most diagnostic tasks, including medical diagnosis as
  of faults in mechanical devices are examples of
well as diagnosis
classification.

Propose and Refine: Many design and planning problems can be
attacked with this strategy.
Heuristic Search Techniques

  and several
A framework for describing search methods is provided
general-purpose search techniques are discussed.

There are many varieties of heuristic search. They can be described
independently of any particular task or problem domain. But when
 highly dependent on
applied to particular problems, their efficacy is
the way they exploit domain-specific knowledge

Over last three decades of AI research, these techniques continue to
provide the framework into which domain-specific knowledge can be
 learning. Thus they
placed, either by hand or as a result of automatic
continue to form the core of most Al systems.
Best-First Search

  of both depth-
Best-first search, is a way of combining the advantages
first and breadth-first search into a single method.

Depth-first search is good because it allows a solution to be found
 
without all competing branches having to be expanded. Breadth-first
search is good because it does not get trapped on dead-end paths.

One way of combining the two is to follow a single path at a time, but
 switch paths whenever some  competing path looks more promising
than the current one does.

At each step of the best-first search process, we select the most
promising of the nodes we have generated so far. This is done by
applying an appropriate heuristic function to each of them, We then
 expand the chosen node by using the rules to generate its successors.

are
If one of them is a solution, we can quit. If not, all those new nodes
added to the set of nodes generated so far. Again the most promising
 Artificial Intelligence Unit-II
node is selected and the process continues. A bit of depth first
searching occurs as the most promising branch is explored.

But eventually, it a solution is not found, that branch will start to look
less promising than one of the top-level branches that had been
ignored. At that point, the now more promising, previously ignored
branch will be explored. But the old branch is not forgotten, Its last
node remains in the set of generated but unexpanded nodes. The
search can return to it whenever all
 the others get bad enough, that it
 is again the most promising path.

Initially, there is only one node, so it will be expanded. Doing so
generates three new nodes. The heuristic function, is an estimate of
the cost of gettingto a solution from a given node, is applied to each of
 these new nodes.

 Since node D is the most promising, it is expanded next, producing two 
successor nodes, E and F. Now the heuristic function is applied to them.

Now another path that going through node B looks more promising, so
it is pursued, generating nodes G and H. But again when these new
nodes are evaluated they look less promising than another path, so
attention is returned tothe path through D to E. E is then expanded,
 yielding nodes I and J.

At the next step, J will be expanded, since it is 
the most promising. This
process can continue until a solution is found.
 Artificial Intelligence Unit-II

In best-first search, one move is selected, but the others are kept
around so that they can be revisited later if the selected path becomes
 less promising.

Further, the best available state is selected in best first search, even it
  that is lower than the value of the state that
that state has a value
was just explored.

 In a best-first search of a tree, it is sometimes important tosearch a
graph instead so that duplicate paths will not be pursued.

An algorithm to do this will operate by searching a directed graph in
which each node represents a point in the problem space. Each node
will contain, in addition to a description of the problem state it
represents, an indication of how promising it is; a parent link that

points back to the best node from  which it came, and a list of the
nodes that were generated from it.

The parent link will make it possible to recover the path to the goal
once the goal is found. The list of successors will make it possible, if a
better path is found to an already  existing node, to propagate the
 improvement down to its successors.

  each of its branches
We will call a graph of this sort an OR graph, since
represents an alternative problem-solving path.

To implement  such a graph-search procedure, we will need to use two
 lists of nodes:
 OPEN—nodes that have been generated and have had the heuristic
function applied to them but which have not yet been examined
(i.e.. had their successors generated) OPEN is actually a priority
queue in which the elements with the highest priority are those
 with the most promising value of the heuristic function.
 CLOSED— -nodes that have already been examined. We need to
keep these nodes in memory if we want to search a graph rather
than a tree, since whenever a new node is generated; we need to
 check whether it has been generated before.

We will also need a heuristic function that estimates the merits of
 each node we generate. The function f1 that gives the true evaluation
of the node.
 
We define this function as the sum of two components: g and h1.

 The function g is ameasure of the cost of getting from the initial state
to the current node.

The function h1 is an estimate of the additional cost of getting from the
 current node to a goal state.

The combined function f1 represents an estimate of the cost of getting 
from the initial state to a goal state along the path that generated the
 Artificial Intelligence Unit-II
current node. If more than one path is generated the node, then the
algorithm will record the best one.
  This process can be summarized as follows:
 Algorithm: Best-First Search
1. Start with OPEN containing just the initial state.
2. Until a goal is found or there are no nodes left on OPEN do:
(a) Pick the best node on OPEN.
(b) Generate its successors.
(c) For each successor do:
i. If it has not been generated before, evaluate it, add it to
OPEN, and record its parent.
ii. If it has been generated before, change the parent if this new
path is, better than the previous one. In that case, update the
cost of getting to this node and to any successors that this node
may already have.
  
The best first search algorithm is a simplification of an algorithm called
 A* .

 same f1, g, and h1 functions, as well as the lists
This algorithm uses the
OPEN and CLOSED.
Iterative Deepening:

A number of ideas for searching two-player game trees have led to new
algorithm for single-agent heuristic search.  One such idea is
 “iterative deepening” originally used in CHESS.

Rather than searching to a fixed depth in the game tree, CHESS first
 searched only a single ply, applying its static evaluation function to
the result of each of its possible moves.

  This was
It then initiated a new minimax search, to a depth of two ply.
followed by a three-ply search, then a four-ply search etc.

 The name "iterative deepening" derives from the fact that on each
iteration, the tree is searched one level deeper.

Since it is impossible to know in advance how long a fixed-depth tree
search will take, a program may find itself running out of time. With
iterative deepening, the current search can be aborted at anytime and
 the best move found by the previous iteration can be played.

can also be used in improve the performance of the A*
Iterative deepening
search algorithm.

 Algorithm: Iterative-Deepening-A*
1. Set THRESHOLD = the heuristic evaluation of the start state.
2. Conduct a depth-first search, pruning any branch when its total
cost function (g + h1) exceeds THRFSHOLD. If a solution path is found
during the search, return it.
 Artificial Intelligence Unit-II
3. Otherwise, increment THRESHOLD by the minimum amount it was
exceeded during the previous step, and then go to Step 2.


Iterative-Deepening-A* (IDA*) is guaranteed to find an optimal solution
provided that h1 is an admissible heuristic.
Problem Reduction: AND-OR Graphs

AND-OR graph (or tree), is useful for representing the solution of
 problems that can be solved by decomposing them  into a set of
smaller problems, all of which must then he solved.

This decomposition, or reduction, generates arcs that we call AND
 arcs. One AND arc may point to any number of successor nodes all of
which must be solved in order for the arc to point to a solution.

Several arcs may emerge from a single node, indicating a variety of ways
in which the original problem might be solved. This is why the structure
is called not simply an AND graph but rather an AND-OR  graph. AND
arcs are indicated with a line connecting all the components.


In order to find solutions in an AND-OR graph, we need an algorithm
with the ability to handle the AND arcs appropriately. This algorithm

should find a path from the starting node of the graph to a set of
nodes representing solution states.

The top node, A, has been expanded, producing two arcs, one leading to
B and one leading to C and D. The numbers at each node represent the
value of f1 at that node. Assume, that every operation has a uniform

cost, so each arc with a single successor has a cost of I and each AND 
arc with multiple successors has a cost of 1 for each of its components.

 1
 for expansion the one with the
If we look just at the nodes and choose
lowest; f value, we must select C.

But using the information now available, it would be better to explore
the path going through B since to use C we must also use D. (for a
 9 (C+D+2) compared to the cost of 6 that we get by going
total cost of
through B.

The problem is that the choice of which node to expand next must
1

depend not only on the f value of that node but also on whether that
node is part of the current best path from the initial node.
 Artificial Intelligence Unit-II

 1
The most promising single node is (G with an f value of 3). It is even
part of the most promising arc G-H, with a total cost of 9. But that are

is not part of the current best path since to use it we must also use
the arc I-J with a cost of 27.

The path from A through B, to E and F is better, with a total cost of
18. So we should not expand G next: rather we should examine either
 E or F.

In order to describe an algorithm for searching an AND-OR graph, we
need to exploit a value called FULILITY. If the estimated cost of a
 
solution becomes greater than the value of FUTILITY, then we
abandon the search.

FUTILITY should be chosen to correspond to a threshold such that any 
solution with a cost above it is too expensive to be practical, even if it is
found.

1 Initialize the graph to the starting node.

2. Loop until the starling node is labelled SOLVED or until its cost goes
above FUTILITY:
(a) Traverse the graph, starting at the initial node and following the
current best path, and accumulate the set of nodes that are or that
path and not yet been expanded or labelled as solved.
b) Pick one of these unexpanded nodes and expand it. If there are no
successors, assign FUTILITY as the value of this node, otherwise, add
1 1
its successors to the graph and for each of them compute f .If f of
any node is 0, mark that node as SOLVED.
(c) Change the f1 estimate of the newly expanded node to reflect the new
information provided by its successors. Propagate this change backward
through the graph. If any node contains a successor arc whose
descendants are all solved, label the node itself as SOLVED. At each node
that is visited while going up the graph, decide which of its successor
arcs is the most promising and mark it as part of the current
 Artificial Intelligence Unit-II
best path. But now expanded nodes must be re-examined so that the
1
best current path can he selected. Thus it is important that their f
values be the best estimates available.


At step 1, A is the only node, so it is at the end of the current best
path. It is expanded, yielding nodes B, C and D. The arc to D is
labelled as the most promising one emerging from A, since it costs 6
compared to B and C, which costs 9. (Marked arcs are indicated in the
 figures by arrows.)

In step 2, node D is chosen for expansion. This process produces one
 new arc, the AND arc to E and F, with a combined cost estimate of 10.
So we update the f1 value of D to 10.

 Going back one more level, we can see that this makes the AND arc  B-
C better than the arc to D, so it is labelled as the current best path.

At step 3, we traverse that arc from A and discover the unexpanded
nodes B and C. If we want to find a solution along this path, we will
have to expand both B and C eventually, so let's choose  to explore B
 first. This generates two new arcs, the ones to G and to H.

 Propagating their f1 values backward, we update f1 of B to 6. 
This
requires updating the cost of the AND arc B-C to 12 (6+4+2).

After doing that, the arc to D is again the better path from A, so we
 record  and either node E or node F will
that as the current best path
be chosen for expansion at step 4.
 
This process continues until either a solution is found or all paths have
led to dead ends, indicating that there is no solution.

  The problem reduction algorithm is a simplification of an AO* algorithm.


The AO* algorithm will use a single structure GRAPH, representing the
part of the search graph that has been explicitly generated so far.
 Artificial Intelligence Unit-II

Each node in the graph will point both down to its immediate successors
and up to its immediate predecessors. Each node in the graph will also
have associated with it an h1 value, an estimate of the  cost of a path
from itself to a set of solution nodes. So h1 will serve
as the estimate of goodness of a node.
Algorithm: AO*
1. Let GRAPH consists of only the node representing the initial state.
(Call this node INIT). Compute h1 (INIT).
2. Until INIT is labelled SOLVED or until INIT’s h1 value becomes greater
than FUTILITY, repeat the following procedure:
(a) Trace the labelled arcs from INIT and select for expansion one of
the yet unexpanded nodes that occurs on this path. Call the selected
node NODE.
(b) Generate the successors of NODE. If there are none, then assign
FUTILITY as the h1 value of NODE. 'This is equivalent to saying that
NODE is not solvable. If there are successors, then for each one called
SUCCESSOR that is not also an ancestor of NODE do the following:
i. Add SUCCESSOR to GRAPH.
ii. If SUCCESSOR is a terminal node, label it SOLVED and
assign it, an h1 value of 0.
iii. If SUCCESSOR is not a terminal node, compute its h1 value.
(c) Propagate the newly discovered information up the graph by doing
the following: Let S the a set of nodes that have been labelled SOLVED
or whose h1 values have been changed and so need to have values
propagated back to their parents. Initialize S to NODE. Until S is
empty, repeat the following procedure:
i. If possible, select from S a node none of whose descendants in
GRAPH occurs in S. If there is no such node, select any node
from S. Call this node CURRENT, and remove it from S.
ii. Compute the cost of each of the arcs emerging from CURRENT.
The cost of each arc is equal to the sum of the h1' values of each of
the nodes at the end of the arc plus whatever the cost of the arc
itself is. Assign as CURRENT's new h1 value the minimum or the
costs just computed for the arcs emerging from it.
iii. Mark the best path out of CURRENT by marking the arc that
had the minimum cost as computed in the previous step.
iv. Mark CURRENT SOLVED if all of the nodes connected to it
through the new labelled arc have been labelled SOLVED.
v. If CURRENT has been labelled SOLVED or if the cost of
CURRENT was just changed, then its new status must be
propagated back up the graph. So add all of the ancestors of
CURRENT to S.
 Artificial Intelligence Unit-II

Constraint Satisfaction

Many problems in Al can be viewed as problems of Constraint
 Satisfaction in which goal is to  discover some problem state that
satisfies a given set of constraints.
  Examples of this sort of problem include crypt arithmetic puzzles.

Design tasks can also be viewed as Constraint Satisfaction problems
  must be created within fixed limit, on time, cost,
in which a design
and materials.

By viewing a problem as one of constraint satisfaction, its often
 
possible to reduce substantially the amount of search that, is required
as compared with a method that attempts to form partial solutions.

A constraint satisfaction approach to solving this problem avoids
making guesses on particular assignments of numbers to letters.
Instead, the initial set of constraints, which says that each number
may correspond to only one letter and that the sums of the digits
must be as they are given in the problem, is first augmented  to
 include restrictions that can be inferred from the rules of arithmetic.

  may still be required, the number of allowable
Then, although guessing
guesses is reduced.

 
Constraint satisfaction is a search procedure that operates in a space of
Constraint sets.

The initial state contains the constraints that are originally given in the
problem description. A goal state is any state that has been constrained -
 enough." where "enough' must be defined for cacti problem.

Constraint satisfaction is a two-step process. First, constraints are
discovered and propagated as far as possible throughout the system.
Then, if there is still not a solution, search begins. A guess about

something is made and added as a new constraint.  Propagation can
then occur with this new constraint, and so forth.

The first step, propagation, arises from the fact that there are usually
dependencies among the constraints. These dependencies occur
because many constraints involve more than  one object and many
 objects participate in more than one constraint.

So, for example, assume we start with one constraint: N = E + I. Then,
 if we added the constraint N = 3, we could propagate that to get a
stronger constraint on E, namely that E = 2.
  Constraint propagation terminates for one of two reasons.
 First, a contradiction may be detected. If this happens, then there
is no solution consistent with all the known constraints. If the
contradiction involves only those constraints that were given as
part of the problem specification (as opposed to ones that were
guessed during problem solving), then no solution exists.
 Artificial Intelligence Unit-II

 The second possible reason for termination is that the propagation

has run out of steam and there are no further changes that can be
made on the basis of current knowledge. If this happens and a
solution has not yet been adequately specified, then search is
 necessary to get the process moving again.
 At this point, the second step begins. Some hypothesis about a way to
strengthen the constraints must be made. ln case of the crypt
arithmetic problem, for example this usually means guessing a
particular value for some letter. Once this has been done, constraint
propagation can begin again from this new state. If a solution is
found, it can be reported. If still more guesses are required, they can
be made. If a contradiction is detected, then backtracking can he
used to try a different guess and proceed with it.

1. Propagate available constraints. To do this, first set OPEN to the set of all
objects that must have values assigned to them in a complete solution. Then
do until an inconsistency is detected or until OPEN is empty:
(a) Select an object OB from OPEN. Strengthen as much as possible the
set of constraints that apply to OB.
(b) If this set is different from the set that was assigned the last time OB
was examined or if this is the first time OB has been examined, then add
to 0PEN all objects that share any constraints with OB.
(c) Remove OB from OPEN.
2. If the union of the Constraints discovered above defines a solution,
then quit and report the solution.
3. If the union of the constraints discovered above defines a contradiction
then return failure.
4. If neither of the above occurs, then it is necessary to make a guess at
something in order to proceed. To do this, loop until a solution is found
or all possible solutions have been eliminated:
(a) Select an object whose value is not yet determined and select a way
of strengthening the constraints on that object.
(b) Recursively inv006Fke constraint satisfaction with the current set of
constraints augmented by the strengthening constraint just selected.
 Artificial Intelligence Unit-II


 
The goal slate is problem state in which all letters have been assigned a
digit in such a way that all the initial constraints are satisfied.

  proceeds in cycles. At each cycle, two significant
The solution process
things are done :
 Constraints are propagated by using rules that correspond to
 the properties of arithmetic.
 A value is guessed for some letter whose value is not yet
 determined.

In the first step, it does not usually matter a great deal what order the
propagation is done in, since all available propagations will be
 performed before the step ends.

 In the second step, the order in which guesses are tried may have a
substantial impact on the degree of search that is necessary.

A few useful heuristics can help to select the best guess to try first.
For example, if there is a letter that has only two possible values and

other with possible values, there is a better chance of guessing right
on the first than on the second.

Another useful heuristic is that if there is a letter that participates in
many constraints then it is a good idea to prefer it to a letter that
participates in a few. A guess on such a highly constrained letter will
usually lead quickly either to a contradiction (if it’s wrong) or to the
generation of many additional constraints (if it is right). A guess on a
 less constrained letter, on the other hand, provides less information.

Let Cl, C2, C3, and C4 indicate the carry bits out of the columns,
numbering from the right. Initially, rules forpropagating constraints
generate the following additional constraints:
 Artificial Intelligence Unit-II
• M 1, since two single-digit numbers plus a carry cannot total more
than 19; S=8 or 9, since S+M+C3>9 (to generate the carry) and M=1,
S+l+C3>9. So S + C3 > 8 and C3 is at most 1.
• O=0, since S + M(l)+C3(<= 1) must be at least 10 to generate a carry
and it can be at most 11. But M is already 1. So O must be 0.
• N = E or E+1, depending on the value of C2. But N cannot have the
same value as E. So, N=E+l and C2 is 1.
• In order for C2 to be1, the sum of N+ R+CI must be greater than 9,
so N+ R must be greater than 18.
• N + R cannot be greater than 18, even with a carry in, so E cannot
be 9.

At this point, let us assume that no more constraints can be generated.
Then, to make progress from here, we must guess. Suppose E is assigned
the value 2. (We chose to guess a value for E because it occurs three
 Now the next cycle
times and thus interacts highly with the other letters.)
begins. The constraint propagator now observes that:
• N =3, since N = E+ 1.
• R= 8 o 9, s ince R+ N (3)+C1 (1 or 0)=2 or 12. But since N is already
3, the sum of these non negative numbers cannot be less than 3.
Thus R + 3 + (0 or 1)= 12 and R = 8 or 9.
• 2 + D= Y or 2 + D = 10 + Y, from the sum in the rightmost column.
Game Playing

Charles Babbage, the nineteenth-century computer architect, thought
 about programming his Analytical Engine  to play chess and later of
building a machine to play tic-tac-toe.

 Claude Shannon wrote a paper in which he described  mechanisms
that could be used in a program to play chess.
  Alan Turing described a chess-playing program, but he never built it.

Arthur Samuel, succeeded in building the first significant,
operational game-playing program. His program played checkers and,
in addition to simply playing 
the game, could learn from its mistakes
 and improve its performance.

 
There were two reasons that games appeared to be a good domain in
which to explore machine intelligence:

 
They provide a structured task in which it is very easy to measure
success or failure.

They did not obviously require large amounts of Knowledge. They were
thought to be solvable by straightforward search from the starting
 state to a winning position.

A program that simply does a straightforward search of the game tree
will not be able to select even its first move during the lifetime 
of its
opponent. Some kind of heuristic search procedure is necessary.
 Artificial Intelligence Unit-II

We use Generate-and-Test procedures in which the testing is done after
 varying amounts of work by the generator.
 At one extreme, the generator generates entire proposed solutions,
 which the tester then evaluates.
 At the other extreme, the generator generates individual moves in
the search space, each of which is then evaluated by the tester and
 the most promising one is chosen.

To improve the effectiveness of a search-based problem-solving
 program, two things can be done:
 Improve the generate procedure so that only good moves (or paths)
 are generated.
 Improve the test procedure so that the best moves (or paths) will be
 recognized and explored first.

The ideal way to use a search procedure to find a solution to a
 
problem is to generate moves through the problem space until a goal
state is reached.

In order to choose the best move, the resulting board positions must
be compared to discover which is most advantageous. This is done

using a static evaluation function, which uses whatever information
it has to evaluate so that the best next move can be chosen.
The Minimax Search Procedure

 The minimax search procedure is a depth-first, depth-limited search
procedure.

 
The idea is to start at the current position and use the plausible-move
generator to generate the set of possible successor positions.

We apply the static evaluation function to those positions and simply
choose the best one. Then we can back  that value up to the starting
 position to represent our evaluation of it.

 Our goal is to maximize the value of the static evaluation function of
the next board position.

Assume a static evaluation function that returns values ranging from -
10 to 10, with 10 indicating awin for us, and -10 a win for the
 opponent, and 0 an even match.

Since our goal is to maximize the value of the heuristic function, we
choose to move to B. Backing B's value up to A, we can conclude that
A’s value is 8.
 Artificial Intelligence Unit-II


After our move, the situation would appear to be very good. But, if we
 look one move ahead, we will see that one of our pieces also  gets
captured and so the situation is not as favourable as it seemed.

So we would like to look ahead to see what will happen to each of the
new gamepositions at the next move which will be made by the
 opponent.

Instead of applying the static evaluation function to each of the
positions that we just generated, we apply the plausible-move 
generator, generating a set of successor positions for each position.
 Artificial Intelligence Unit-II


But now we must take into account that the opponent gets to choose
 which successor moves to make and  thus which terminal value
should he backed up to the next level.

Suppose we made move B. Then the opponent must choose among
moves E, F, and G. The opponents goal is to minimize the value of the
evaluation function, so he or she can be expected to choose move F.
This means that if we make move B, the actual position in which we
will end up one move later is very bad for us. If node E is selected, it is
very good for us. Since  at this level we are not the ones to move, we
 will not get to choose it.

At the opponent's choice, the minimum value was chosen and backed
up. At the level representing our choice, the maximum value was
 chosen.

Once the values from the second ply are backed up, it becomes clear
that the correct move for us to make at the first level, given the
  the opponent can do from
information, is C, since there is nothing
there to produce a value worse than –2.

This process can he repeated for as many ply as time follows, and
more accurate evaluations that  are produced can be used to choose
 the correct move at the top level.

The alternation of maximizing and minimizing at alternate ply when

evaluations are being pushed back up corresponds to the opposing 
strategies of the two players and hence this method is called. minimax.

 The recursive procedure that relieson two auxiliary procedures that are
specific to the game being played:
 MOVEGEN(Position,Player) -The plausible-move generator, which
returns the list of nodes representing the moves that can be made
by player in Position. We call the two players PLAYER-ONE and
PLAYER-TWO. In a chess program, we might use the names
BLACK and WRITE instead.
 Artificial Intelligence Unit-II

 STATIC(Position,Player)—The static evaluation function, which

returns a number representing the goodness of position from the
 standpoint of Player.

As with any recursive program. a critical issue in the design of the
 MINIMAX procedure is when  to stop the recursion and simply call the
static evaluation function.

  are the factors that may influence this decision. They
The following
include:
• Has one side Won?
• How many ply have we already explored?
• How promising is this path?
• How much time is left?
• How stable is the configuration?

For the general MINIMAX, we use a function DEEP-ENOUGH, which is
 assumed to evaluate all of these factors and to return TRUE if the 
search should be stopped at the current level and FALSE otherwise.

 DEEP-ENOUGH will take two parameters:
The implementation of
Position and Depth.

 return TRUE if its Depth
It will ignore its Position parameter and simply
parameter exceeds a constant cut off value.
  M1NIMAX as a recursive procedure needs to return two results:

  The backed-up value of the path it chooses.

 The path itself. We return the entire path even though probably
only the first element, representing the best move from the current
position, is actually needed.
  MINIMAX returns a structure containing both results: VALUE and 
 PATH.

 We define the MININMAX procedure as  a recursive function, called
initially that takes three parameters.
  A board position.
  The current depth of the search, and
  The player to move.

 call to compute the best move from the position CURRENT
So the initial
should be :
  MI1N1MAX(CURRENT, O PLAYER-ONE) if PLAYER-ONE is to move,
 MINIMAX (CURRENT, 0, PLAYER-TWO) if PLAYER TWO is to move.

Algorithm: MINIMAX (Position, Depth, Player)

1. If DEEP-ENOUGH (Position, Depth) then return the
structure. VALUE= STATIC (Position, Player);
PATH = nil
 Artificial Intelligence Unit-II
This indicates that there is no path from this node and that its value
is that determined by the static evaluation function,
2. Otherwise, generate one more ply of the tree by calling the function
MOVEGEN (Position, Player) and setting SUCCESSORS to the list it
returns.
3. If SUCCESSORS is empty, then there are no moves to be made, so
return the same structure that would have been returned it DEEP-
ENOUGH had returned true.
4. If SUCCESSORS is not empty, then examine each element in turn
and keep track of the best one. This is done as follows.
Initialize BEST-SCORE to the minimum value that STATIC can return
It will be updated in reflect the best score that can be achieved by an
element of SUCCESSORS.
For each element SUCC of SUCCESSORS, do the
following: (a)Set RESULT-SUCC to
MINIMAX(SUCC, Depth + i, OPPOSITE(Player))
(b)Set NEW_VALUE to –VALUE (RESULT-SUCC).
(c) If NFW-VALUE > BEST-SCORE, then we have found a successor
that better than any that have been examined so far. Record this by
doing the following:
i. Set BEST-SCORE to NEW-VALUE.
ii. The best known path is now from CURRENT to SUCC and
then on to the appropriate path down from SUCC as determined
by the recursive call to MINIMAX. So set BEST-PATH to the
result of attaching SUCC to the front of PATH(RESUIT-SUCC);
5. Now that all the successors have been examined, we know the value of
Position as well as which Path to take from it. So return the structure:
VALUE = BEST-SCORE
PATH = BEST- PATH
Alpha-Beta Cutoffs

The minimax procedure is a depth-first process. One path is explored
as far as time allows, the static evaluation function is applied to the
  and the value can then be
game positions at the last step of the path,
passed up the path one level at a time.

The efficiency of depth-first procedures can be improved by using
branch-and-bound techniques in which partial solutionsthat are
 clearly worse than known solutions can be abandoned early.

  so far. If
For this, we require to store of the length of the best path found
a later partial path outgrew that bound, it was abandoned.

It is necessary to modify
our search procedure to handle both maximizing
and minimizing players.
 Artificial Intelligence Unit-II

 It is also necessary to modify the branch-and-bound strategy to include 
two bounds, one for each of the players. This modified strategy is called
“alpha-beta pruning”. It requires the maintenance of two threshold
 values:
 a lower bound on the value that a maximizing node may ultimately
 be assigned called alpha.
 an upper bound on the value that a minimizing node may be
assigned called beta.


After examining node F, we know that the opponent is guaranteed a
score of —5 or less at C (since the opponent is the minimizing player).
But we also know that we are guaranteed a  score of 3 or greater at
 node A, which we can achieve if we move to B.

Any other move that produces a  score of less than 3 is worse than the
move to B, and we can ignore it.
 Artificial Intelligence Unit-II

After examining only F, we are sure that a move to C is worse (it will
be less than or equal to —5) regardless of  the score of node G. Thus
we need not bother to explore node G at all.


Alpha Value: In searching the tree, the entire sub tree headed by B is
searched, and we discover that at A we can expect a score of at least
  is passed down to F, it will enable us to skip
3. When this alpha value
the exploration of L.

The reason is as follows: After K is examined; we see that I is
guaranteed a maximum score of 0, which means that F is guaranteed

a minimum of 0. But this is less than alpha's value of 3, no more
branches of I need be considered.

The maximizing player already knows not to choose to move to C and
 
then to I since, if that move is made, the resulting score will be no better
than 0 and a score of 3 can be achieved by moving to B instead. Nov,

Beta Value: After cutting off further exploration of I, J is examined
yielding a value of 5, which is assigned as the value of F. This value
becomes the value of beta at node C. It indicates that C is guaranteed
to get a 5 or less.
 Artificial Intelligence Unit-II

Now, we must expand G. First M is examined and it has a value of 7,
which is passed back to G as its tentative value. But now 7 is
compared to beta (5). It is greater, and the player whose turn it is at
node C is trying to minimize. So this player will not choose G, which
would lead to a score of at least 7, since there is in alternative move to
F, which will lead to a score of 5. Thus it is not necessary to explore
 any of the other branches of G.
  The function MINIMAX-A-B, which requires four arguments: Position,
 Depth, Use-Thresh, and Pass-Thresh.

  a move for PLAYER-ONE from the position
The initial call, to choose
CURRENT should be
MINIMAX-A-
 B(CURRENT 0,
 PLAYER-ONE,
maximum value STATIC can compute
 minimum value STATIC can compute)
The initial values for Use-Thresh and Pass-Thresh represent the worst
values that each side could achieve.

Algorithm: MINIMAX-A-B((Position, Depth, Player, Use-Thresh, Pass-

Thresh)
1. If DEEP-ENOUGH (Position, Depth) then return the
structure. VALUE= STATIC (Position, Player);
PATH = nil
This indicates that there is no path from this node and that its value
is that determined by the static evaluation function,
2. Otherwise, generate one more ply of the tree by calling the function
MOVEGEN (Position, Player) and setting SUCCESSORS to the list it
returns.
3. If SUCCESSORS is empty, then there are no moves to be made, so
return the same structure that would have been returned it DEEP-
ENOUGH had returned true.
4. If SUCCESSORS is not empty, then examine each element in turn
and keep track of the best one. This is done as follows.
Initialize BEST-SCORE to the minimum value that STATIC can return
It will be updated in reflect the best score that can be achieved by an
element of SUCCESSORS.
For each element SUCC of SUCCESSORS, do the following:
(a) Set RESULT-SUCC to
MINIMAX-A-B (SUCC, Depth + 1, OPPOSITE(Player),
-Pass-Thresh, -Use-Thresh).
(b) Set NEW-VALUE to —VALUE (RESULT-SUCC).
 Artificial Intelligence Unit-II
(c) If NEW-VALUE > Pass-Thresh, then we have found a successor
that is better than any that have been examined so far. Record this by
doing the following:
i. Set Pass- Thresh to NEW-VALUE.
ii. The best known path is now from CURRENT to SUCC and
then on to the appropriate path from SUCC as determined by
the recursive call to MINIMAX-A-B. So set BEST-PATH to the
result of attaching SUCC to the front of PATH (RESULT-SUCC).
(d) If Pass-Thresh (reflecting the current best value) is not better
than Use-Thresh, then we should stop examining this branch.
But both thresholds and values have been inverted. So if Pass-
Thresh >= (Use-Thresh, then return immediately with the value
VALUE = Pass-Thresh
PATH = BEST-PATH
5. Return the structure:
VALUE Pass-Thresh
PATH = BEST-PATH

Assignment - Cum - Tutorial Questions

I) Objective Questions

1. Define the term “state space”.

2. Define the term “Operationalization”.
3. A ____________ is a technique that improves the efficiency of a search
process.
[ ]
(a) Heuristic (b) Control Strategy (c) GPS (d)
none

4. List the steps for solving the problem.

5. In Theorem Proving, the solution steps can be_________. [ ]
(a) Ignored (b) Recoverable (c) Irrecoverable (d) none

6. Problems in which solution steps cannot be undone are ______. [ ]

(a) Ignored (b) Recoverable (c) Irrecoverable (d) none
7. Travelling Salesman problem is an example of_________. [ ]
(a) Best-Path (b) Any-Path (c) Both (d) none

8. ____________ are the problems in which the computer is given a

problem description and produces an answer with no intermediate
 Artificial Intelligence Unit-II
communication.
[ ]
(a) Conversational (b) Solitary (c) Ignorable(d) None

9. _________is the list of nodes that have been generated and have had
the heuristic function applied to them but which have not yet been
examined.
[ ]
(a) OPEN (b) CLOSED (c) NODES (d) none
10. The function ___ is a measure of the cost of getting from the initial
state to the current node. [ ]

(a) f1 (b) g (c) h1 (d) none

11. ____________ is useful for representing the solution of problems

that can be solved by decomposing them into a set of smaller problems,

all of which must then he solved. [ ] (a) OR graph (b) AND graph (c)
AND-OR graph (d) none
12. ____________ function uses whatever information it has to
evaluate so that the best next move can be chosen. [ ]
(a) static evaluation (b) dynamic evaluation

(c) Threshold evaluation (d) none

13. DEEP-ENOUGH will take two parameters: ________ and ___________.

14. MINIMAX returns a structure containing both results: ________ and
______________.
15. A lower bound on the value that a maximizing node may ultimately be

assigned called __________. [ ] (a) alpha (b) beta (c) gamma (d) none

16. An upper bound on the value that a minimizing node may be assigned

called ____________. [ ] (a) alpha (b) beta (c) gamma (d) n

 Artificial Intelligence Unit-II
II) Descriptive Questions

1. List the steps for General Problem Solving.

2. Give an example of a problem for which breadth -first search would
work better than depth-first search. Give an example of a problem for
which depth-first search would work better than breadth-first search.
3. Describe the state space of water jug problem and also explain its
solution.
4. List the steps necessary to provide a formal description of a problem.
5. What are the requirements of a control strategy? Develop an algorithm
for: (i) Breadth First Search (ii) Depth First Search
6. Summarize the advantages and disadvantages of control strategies.
7. Outline the factors that are necessary for analyzing a problem to
choose most appropriate heuristic method.
8. Explain non-decomposable problem with suitable example.
9. Distinguish between Ignorable, Recoverable and Irrecoverable
problems with necessary examples.
10. Distinguish between:
(i) Certain Outcome Vs Uncertain Outcome problems
(ii) Best path and Any-path problems

(iii) Problems whose solution is a state and whose solution is a path to

state.

(iv) Solitary Vs Conversational problems

11. Explain an algorithm for Best-first Search.
12. Explain Problem Reduction using AND-OR graph with an algorithm.
13. Explain AO* algorithm.
14. Explain the procedure of Constraint Satisfaction with an example.
15. Explain Minimax procedure with an example.
16. Explain Alpha-Beta Pruning with an example.
17. Solve the Criptarithmetic problem using Constraint Satisfaction:
CROSS
ROADS
________
DANGER
_________
 AI 1

UNIT - III: Logic Concepts

Syllabus:

UNIT - III: Logic Concepts

Introduction, propositional calculus, proportional logic, natural deduction
system, axiomatic system, semantic tableau system in proportional logic,
resolution in proportional logic, resolution in predicate logic and unification
algorithm.

3.1 Introduction
  One particular way of representing facts is the language of logic.
 The logical formalism is appealing because it immediately suggests a
powerful way of deriving new knowledge from old called as
 “mathematical deduction”.
 In this formalism, we can conclude that a new statement is true by
proving that it follows from the statements that are already known.
 The way of demonstrating the truth of an already believed proposition
can be extended to include deduction as a way of deriving answers to
 questions and solutions to problem.
 we use the following standard logic
symbols: (Implication)

¬ (Not)
Λ (And)
V (Or)
∃ (there exists) ∀ (For All)

3.2 Propositional logic

 Propositional logic (PL) is the simplest form of logic where all the
statements are made by propositions. A proposition is a declarative
statement which is either true or false. It is a technique of knowledge
 representation in logical and mathematical form.
  Propositional logic is also called Boolean logic as it works on 0 and 1.
 In propositional logic, we use symbolic variables to represent the logic,
and we can use any symbol for a representing a proposition, such A,
 B, C, P, Q, R, etc.
 Propositions can be either true or false, but it cannot be
 AI 2

 Propositional logic consists of an object, relations or function, and

logical connectives.
  These connectives are also called logical operators.
 The propositions and connectives are the basic elements of the
propositional logic.
 Connectives can be said as a logical operator which connects two
sentences.
 A proposition formula which is always true is called tautology, and it
is also called a valid sentence.
  A proposition formula which is always false is called Contradiction.
  A proposition formula which has both true and false values is called
 Statements which are questions, commands, or opinions are not
propositions such as "Where is Rohini", "How are you", "What is your
 name", are not propositions.
 The syntax of propositional logic defines the allowable sentences for
the knowledge representation. There are two types of Propositions:
Atomic Propositions: Atomic propositions are the simple
propositions. It consists of a single proposition symbol. These are
 the sentences which must be either true or false.
 Examples:
a) 2+2 is 4, it is an atomic proposition as it is a true fact.
b) "The Sun is cold" is also a proposition as it is a false fact.
Compound propositions: Compound propositions are constructed
by combining simpler or atomic propositions, using parenthesis
and logical connectives.
Examples:
a) "It is raining today, and street is wet."
b) "Ankit is a doctor, and his clinic is in Mumbai."
 Logical connectives are used to connect two simpler propositions or
representing a sentence logically. We can create compound
propositions with the help of logical connectives. There are mainly five
connectives, which are given as follows:
1. Negation: A sentence such as ¬ P is called negation of P. A literal
can be either Positive literal or negative literal.
2. Conjunction: A sentence which has ∧ connective such as, P ∧ Q is
called a conjunction.
Example: Rohan is intelligent and hardworking. It can be
written as,
 P= Rohan is intelligent,
Q= Rohan is hardworking. → P∧ Q.

3. Disjunction: A sentence which has ∨ connective, such as P ∨ Q. is
called disjunction, where P and Q are the propositions
 AI 3

Example: "Ritika is a doctor or Engineer", Here P= Ritika is Doctor.

Q= Ritika is Doctor, so we can write it as P ∨ Q.

4. Implication: A sentence such as P → Q, is called an implication.

Implications are also known as if-then rules. It can be represented
as:
If it is raining, then the street is wet.
Let P= It is raining, and Q= Street is wet, so it is represented as :
P→Q
5. Biconditional: A sentence such as P⇔ Q is a Biconditional
sentence, example If I am breathing, then I am alive P= I am
breathing, Q= I am alive, it can be represented as P ⇔ Q.
 Truth Tables:

P Q ¬Q P∧Q P∨Q P→Q P⇔ Q

T T F T T T T
T F T F T F F
F T F F T T F
F F T F F T T

  Limitations of Propositional logic:

We cannot represent relations like ALL, some, or none with
propositional logic. Example:
All the girls are intelligent.
Some apples are sweet.
 Propositional logic has limited expressive power.
In propositional logic, we cannot describe statements in terms of
their properties or logical relationships.
3.3 Semantic Tableau in propositional logic
 Semantic tableau is designed to make semantic reasoning fully
systematic and to generate decisions automatically as to whether
 sequents are valid or not.
 The idea of a tableau is to take a set of formulae with labels saying
which are true and which false, and to process them by analysing
them into their subformulae, transferring the labels to those in
accordance with the semantics of the various connectives, until either
 the analysis is complete or a contradiction results.
 Each node of the tableau is characterised by a set of labelled
formulae, and if it has successor nodes, these result by analysis of
some formula in the set into its principal subformulae. For example,
suppose there is a conjunction A ∧ B in the set with the label t (i.e.
true). For a conjunction to be true is exactly for both of its conjuncts
 AI 4

to be true, so we can choose to extend the tableau by analysing the

conjunction, in which case the successor node will have the two
formulae A and B each with label t in place of the conjunction. On the
other hand, if we choose to analyse a disjunction A ∨ B also with the
label t, there will be no unique successor node, since there is nothing
to say which disjunct should be the true one. Instead, there will be
two successors, one with A labelled t in place of the disjunction, and
the other with B.
 The rules for extending tableaux to analyse negations, conjunctions
and disjunctions:

A set of labelled formulae including the one above the line may be
developed by replacing that formula with the one or more formulae
below the line. A vertical bar separating two sets below the line,
indicates that there are two successor nodes, so the branch of the
tableau splits into two at that point.
 A contradiction in a tableau branch causes that branch to die: it gets
closed off and no more growth happens from its tip. If every branch
dies, the tree dies; in that case we know there was no way the tested
formulas could all have had the truth values they were initially
 supposed to have.
 If, on the other hand, we get to a state
in which every subformula has been
processed in every branch in which it
occurs and some branch is left alive or
"open", we have found an
interpretation on which all of the
original formulas are true.
Conventionally we mark a dead or
"closed" branch by writing a cross
under it.
 AI 5

3.4 Representing Simple Facts in Logic

 Propositional logic is appealing because it is simple to deal with and a
decision procedure for it exists.
 We can easily represent real-world facts as logical propositions written
as well formed formulas (wff’s) in Propositional logic.
 Some Simple Facts in Propositional logic:
It is raining
RAINING
It is sunny
SUNNY
It is windy
WINDY
If it is raining, then it is not sunny.
RAINING¬SUNNY
 Suppose we want to represent the fact stated by the
sentence: Socrates is a man
 We could write SOCRATESMAN. It is represented as:
MAN (SOCRATES)
Plato is a man

 Now the structure of the representation reflects the structure of the

knowledge itself. But to do that, we need to be able to use
 “predicates” applied to arguments.
 To represent the following sentence:
All men are mortal
We can’t represent this as: MORTALMAN

But that fails to capture the relationship between any individual

being a man and that individual being a mortal. To do that, we need
variables and quantification.
 So we use first-order predicate logic as a way of representing
knowledge because it permits representations of things that cannot be
 represented in propositional logic.
 In predicate logic, we can represent real-world facts as statements
written as wff's.
 The major motivation for choosing to use logic is that it we use logical
statements as a way of representing knowledge, and then we have a
 good way of reasoning with that knowledge.
 Determining the validity of a proposition in propositional logic is
straightforward, although it may be computationally hard.
 AI 6

 Before we use predicate logic as a medium for representing

knowledge, we need to check whether it also provides a good way of
 reasoning with the knowledge.
 It provides a way of deducing new statements from old ones.
Unfortunately, unlike propositional logic, it does not possess a
decision procedure.
There do exist procedures that will find a proof of a proposed
theorem, but first-order predicate logic is not decidable, it is
 semi-decidable.
A simple such procedure is to use the rules of inference to
generate theorems from the axioms in some orderly fashion.
This method is not particularly efficient, however, and we will
want to try to find a better one.
Use of predicate logic:
 Let’s now explore the use of predicate logic as a way of representing
knowledge by looking at a following example:
 Consider the following set of sentences:
1. Marcus was a man.
2. Marcus was a Pompeian.
3. All Pompeians were Romans.
4. Caesar was a ruler.

6. Everyone is loyal to someone.

7. People only try to assassinate rulers they are not loyal to.
8. Marcus tried to assassinate Caesar.
 The facts described by these sentences can be represented as a set of
wff’s in predicate logic as follows:
1. Marcus was a man.
man (Marcus)
This representation captures the critical fact of Marcus being a man.
It fails to capture some of the information in the English sentence, the
notion of past tense.
2. Marcus was a Pompeian.
Pompeian (Marcus)
3. All Pompeians were Romans.
∀(x): Pompeian(x)  Roman(x)

4. Caesar was a ruler.

ruler (Caesar)
5. All Romans were either loyal to Caesar or hated him.
∀(x): Roman(x) loyalto(x, Caesar) V hate( x , Caesar)

6. Everyone is loyal to someone.

 AI 7
∀(x) : ∃(y): loyalto(x, y)
7. People only try to assassinate rulers they are not loyal to
∀(x) : ∃(y):person(x) ∧ ruler(y) ∧ tryassassinate(x, y) ¬loyalto(x, y)

8. Marcus tried to assassinate Caesar

tryassassinate(Marcus, Caesar)
 Now suppose that we want to use these statements to answer the
 question:
 Was Marcus loyal to Caesar?
Let's try to produce a formal proof, reasoning backward from the
 desired goal:
¬loyalto (Marcus, Caesar)
 To prove this, let us add a fact:
 All men are people
Although we know that Marcus was a man, we do not have any way to
conclude that Marcus was a person. So, we need the representation

of another fact to our system,

∀(x):

man(x)person(x)

3.5 Representing Instance and isa Relationships

 The specific attributes instance and isa play an important role in the
 useful form of reasoning, property inheritance.
  Let us represent the following facts in three ways:
1. Marcus was a man.
2. Marcus was a Pompeian.
3. All Pompeians were Romans.
 AI 8

4. Caesar was a ruler.

5. All Romans were either loyal to Caesar or hated him.

 Second representation: The predicate instance is a binary one, whose

first argument is an object and whose second argument is a class to
which the object belongs. But these representations do not use an
 explicit isa predicate.
 In subclass relationships, such as that between Pompeians and
Romans, the implication rule there states that it an object is an
instance of the subclass Pompeian then it is an instance of the
 superclass Roman.
 This rule is equivalent to the standard set-theoretic definition of the
subclass-superclass relationship.
 Third representation: It contains representations that use both the
 instance and isa predicates explicitly.
 The use of the isa predicate simplifies the representation of sentence
3, but it requires that one additional axiom. This additional axiom
describes how an instance relation and an isa relation can be
combined to derive a new instance relation.

3.6 Representing facts in Propositional Logic:

Consider the following set of facts:
1. Marcus was a man.
2. Marcus was a Pompeian.
 AI 9

3. Marcus was boni in 40 A.D.

4. All men are mortal.
5. All Pompeians died when the volcano erupted in 79 AD.
6. No mortal lives longer than 150 years.
7. It is now 1991.
8. Alive means not dead.
9. If someone dies, then he is dead at all later times.
Now let's attempt to answer the question:
"Is Marcus alive?"
By proving:
¬ alive(Marcus, now)
The given facts are represented as follows:
 AI 10

3.7 Resolution:
 Resolution is such a procedure, which gains its efficiency from the
fact that it operates on statements that have been converted to a very
convenient standard form.
 Resolution produces proofs by refutation. To prove a statement (to
show that it is valid), resolution attempts to show that the negation of
the statement produces a contradiction with the known statements
 (i.e.. that it is unsatisfiable).
 Resolution is a technique for proving theorems in the propositional or
predicate calculus that has been a part of AI problem-solving.
 Resolution describes a way of finding contradictions in a database of
clauses with minimum use of substitution. Resolution refutation
proves a theorem by negating the statement to be proved and adding
this negated goal to the set of axioms that are known (have been
 assumed) to be true.
 It then uses the resolution rule of inference to show that this leads to
a contradiction. Once the theorem prover shows that the negated goal
is inconsistent with the given set of axioms, it follows that the original
goal must be consistent. This proves the theorem.
 AI 11

  Resolution refutation proofs involve the following steps:

1. Put the premises or axioms into clause form.
2. Add the negation of what is to be proved, in clause form, to the set
of axioms.
3. Resolve these clauses together, producing new clauses that logically
follow from them.
4. Produce a contradiction by generating the empty clause.
5. The substitutions used to produce the empty clause are those
under which the opposite of the negated goal is true.

3.8 Conversion to Clause Form:

 Consider the following fact:
All Romans who know Marcus either hate Caesar or think
that anyone who hates anyone is crazy.
∀x:[Roman(X)⋀know(x,Marcus)][hate(x,Caesar)V(∀y:∃z:hate(y, z)thinkcrazy(x, y))]
Algorithm: Conversion to clause form:
1. Eliminate, using the fact that ab is equivalent to¬ aVb. Performing
this transformation, we get:
∀x:¬[Roman(X)⋀know(x,Marcus)]V[hate(x,Caesar)V(∀y:¬(∃z:hate(y, z))Vthinkcrazy(x, y))]
2. Reduce the scope of each ¬ to a single term, using the fact that ¬(¬P)=P and deMorgan's laws [which say that (¬(a⋀b)=
¬aV ¬b and (¬(a V b)= ¬a⋀ ¬b) and the standard correspondences between quantifiers:- ∀x: P(x)= ∃x: ¬P(x) and ∃x: P(x)= ∀x:
¬P(x) Performing this , we get:
∀x:[¬Roman(X)V¬know(x,Marcus)]V[hate(x,Caesar)V(∀y:∀z:¬hate(y, z))V(thinkcrazy(x, y))]
3. Standardize variables so that each quantifier binds a unique variable.
Since variables are just dummy names, this process cannot affect the truth
value of the wff. For example, the formula:
∀x : P(x) V ∀x: Q(x) would be converted to:
∀x : P(x) V ∀y: Q(y)

4. Move all quantifiers to the left of the formula without changing their
relative order. This is possible since there is no conflict among variable
names. Performing this, we get:
∀x:∀y:∀z:[¬Roman(X)V¬know(x,Marcus)]V[hate(x,Caesar)V(¬hate(y, z)Vthinkcrazy(x, y))]
At this point, the formula is known as Prenex normal form. It consists of a
prefix of quantifiers followed by a matrix, which is quantifier-free.
 AI 12

5. Eliminate existential quantifiers.

For example, the formula
∃y:President(y) can be transformed into the formula:

President(S1) where S1 is a function with no arguments that somehow

produces a value that satisfies President.
If existential quantifiers occur within the scope of universal quantifiers, then
the value that satisfies the predicate may depend on the values of the
universally quantified variables. For example:
∀x: ∃y: father-of(y, x)

The value of' y that satisfies father-of depends on the particular value of x,
we must generate functions with the same number of arguments as the
number of universal quantifiers in whose scope the expression occurs. So
this would be transformed into:
∀x: father-of(S2(x), x))

These generated functions are called Skolem Functions. Sometimes ones

with no arguments are called Skomlem constants.
6. Drop the prefix. All remaining variables are universally quantifled, so the
prefix can just be dropped and any proof procedure we use can simply
assume that any variable it sees is universally quantified.
[¬Roman(X)V¬know(x,Marcus)]V[hate(x,Caesar)V
(¬hate(y, z)Vthinkcrazy(x, y))]
7. Convert the matrix into a conjunction of disjuncts. Since there are no and's, it is only necessary
to exploit the associative property of (i.e., a⋁(b ⋁c) =(a ⋁b)⋁c and simply remove the parentheses.
¬Roman(X)V¬know(x,Marcus)Vhate(x,Caesar)V (¬hate(y, z)
Vthinkcrazy(x, y)
8 Create a separate clause corresponding to each conjunct. In order for a wff
to be true, all the clauses that are generated from it must be true. If we are
going to be working with several wff's, all the clauses generated by each of
them can now be combined to represent the same set of facts as were
represented by the original wff's.
9. Standardize apart the variables in the set of clauses.
(∀x:P(x)⋀Q(x))=∀x:P(x)⋀ ∀x:Q(x)

Since each clause is a separate conjunct and since all the variables are
universally quantified, there is no relationship between the variables of two
clauses, even if they were generated from the same wff.
3.9 The Basis of Resolution
 The resolution procedure is a simple iterative process: at each step
two clauses, called the parent clauses, are compared (resolved),
yielding a new clause that has been inferred from them. The new
 AI 13

clause represents ways that the two parent clauses interact with each
other. Suppose there are two clauses in the system:
winter V summer
¬winter V cold
 Now we observe that precisely one of winter and ¬winter will be true
at any point.
 If winter is true, then cold must be true to guarantee the truth of the
 second clause.
 If ¬winter is true, then summer must be true to guarantee the truth of
the first clause.
 Thus we see that from these two clauses we can deduce:
summer V cold
 Resolution operates by taking two clauses that each contains the
same literal, in this example, winter. The literal must occur in positive
form in one clause and in negative form in the other. The resolvent is
obtained by combining all of the literals of the two parent clauses
 except the ones that cancel.
 If the clause that is produced is the empty clause, then a
contradiction has been found. For example. the two clauses:
Winter
¬winter
will produce the empty clause.
 In predicate logic, the situation is more complicated since we must
consider all possible ways of substituting values for the variables.

3.10 Resolution in Propositional Logic

 In propositional logic, the procedure for producing a proof by
resolution of proposition P with respect to a set of axioms F is the
following
Algorithm: Propositional Resolution
1. Convert all the propositions of F to clause form.
2. Negate P and convert the result to clause form. Add it to the set of clauses
obtained in step 1.
3. Repeat until either a contradiction is found or no progress can be made:
(a) Select two clauses. Call these the parent clauses.
(b) Resolve them together. The resulting clause, called the resolvent,
will be the disjunction of all of the literals of both of the parent clauses
with the following exception: It there are any pairs of literals L and ¬L
such that one of the parent clauses contains L and the other contains
¬L ,then select one such pair and eliminate both L and ¬L from the
resolvent
 AI 14

(c) If the resolvent is the empty clause, then a contradiction has been
found. If it is not, then add it to the set of clauses available to the
procedure.
 Suppose we are given the axioms shown below and we want to prove
R. First we convert the axioms to clause form. Then we negate R.
 producing ¬R, which is already in clause form.
 Then we begin selecting pairs of clauses to resolve together. Although
any pair of clauses can be resolved, only those pairs that contain
complementary literals will produce a resolvent that is likely to lead to
 the goal of producing the empty clause.
 We begin by resolving with the clause ¬R .

 One way of viewing the resolution process is that it takes a set of
clauses that are all assumed to be true and, based on information
provided by the others, it generates new clauses that represent
restrictions on the way each of those original clauses can be made
 true.
 A contradiction occurs when a clause when there is no way it can be
true. This is indicated by the generation of the empty clause.
 AI 15

3.11 The Unification Algorithm

 In predicate logic, this matching process is more complicated since
 the arguments of the predicates must be considered.
 For example, man(John) and ¬man(John) is a contradiction, while
man(John) and ¬man{Spot) is not.
 Thus, in order to determine contradictions, we need a matching
procedure that compares two literals and discovers whether there
exists a set of substitutions that makes them identical. The
straightforward recursive procedure, called the “unification
 algorithm” does this.
 Basic idea of unification: It is very simple. To attempt to unify two
literals, we first check if their initial predicate symbols are the same. If
so, we can proceed. Otherwise, there is no way they can he unified,
 regardless of their arguments.
 For example, the two literals:
trytoassassinate( Marcus, Caesar)
hate(Marcus, Caesar)
cannot be unified.
 If the predicate symbols match, then we must check the arguments
one pair at a time. If the first matches, we can continue with the
second, and so on. To test each argument pair, we can simply call the
unification procedure recursively.
 The matching rules are simple: Different constants or predicates
cannot match; identical ones can. A variable can match another
variable, any constant, or a predicate expression with the restriction
that the predicate expression must not contain any instances of the
 variable being matched.
 We must find a single, consistent substitution for the entire literal,
not separate ones for each piece of it. To do this, we must take each
substitution that we find and apply it to the remainder of the literals
before we continue trying to unify them.
 For example, suppose we want to unify the expressions

 The two instances of P match. Next we compare x and y, and decide

that if we substitute y for x, they could match. We will write that
 substitution as: y/x.
 But now, if we simply continue and match x and z, we produce the
substitution z/x. But we cannot substitute both y and z for x. The
problem can be solved as follows:
 AI 16

 What we need to do after finding the first substitution y/x is to make

that substitution throughout the literals, giving:

 Now we can attempt to unify arguments y and z, which succeeds with

the substitution z/y. The entire unification process has now
succeeded with a substitution that is the composition of the two
 substitutions: (z/y) (y/x).
 In general, substitutions: (a1/a2, a3/a4,....)(b1/b2, b3/b4....) means
to apply all the substitutions of the right most list, then take the
result and apply all the ones of the next list, and so forth, until all
 substitutions have been applied.
 The objective of the unification procedure is to discover at least one
substitution that causes two literals to match.
 For example, the literals:
hate (x, y)
hate (Marcus, z)
could be unified with any of the following substitutions:
(Marcus/x, z/y)
(Marcus/x, y/z)
(Marcus/x, Caesar/y, Caesar/z)
(Marcus/x, Polonius/y, Polonius/z)
 We describe a procedure Unify (Ll, L2), which returns as its value a
list representing the composition of the substitutions that were
performed during the match. The empty list, NIL, indicates that a
match was found without any substitutions. The list consisting of the
single value FAIL indicates that the unification procedure
failed. Algorithm: Unify (L1, L2)
1. If L1 and L2 are both variables and constants, then:
(a) If L1 and L2 are identical, then return NIL.
(b) Else if L1 is a variable, then if LI occurs in L2 then return {FAIL},
else return (L2/L1).
(c) Else if L2 is a variable then if L2 occurs in L1 then return {FAIL},
else return (LI/L2)
(d) Else return {FAIL}.
2. If the initial predicate symbols in LI and L2 are not identical, then return
{FAIL}.
3. If Ll and L2 have a different number of arguments, then return {FAlL}.
4. Set SUBST to NIL. (At the end of this procedure, SUBST will contain all
the substitutions used to unify L1 and L2).
5. For i1 to number of arguments in LI:
 AI 17

(a) Call Unify with the ith argument of LI and ith argument of L2,
putting result in S.
(b) If contains FAIL, then return {FAIL}.
(c) If S not equal to NIL then:
i. Apply S to the remainder of both L1 and L2.
ii. SUBST= APPEND(S, SUBST).
6. Return SUBST.

3.12 Resolution in Predicate Logic

 With Unification, we now have an easy way of determining that two
literals are contradictory, if one of them can be unified with the
negation of the other.
 So, for example, man(x) and ¬man(Spot) are contradictory, since
 man(x) and man(Spot) can be unified, with substitution x/spot.
 In order to use resolution for expressions in the predicate logic, we
use the unification algorithm to locate pairs of literals that cancel out.
 For example, suppose we want to resolve two clauses:
man(Marcus)
¬man(x1) V mortal(x1)
The literal man(Marcus)can be unified with the literal ¬man(x1) with
the substitution Marcus/x1. we can now conclude only that
mortal(Marcus) must be true.
 So the resolvent generated by clauses 1 and 2 must be mortal
(Marcus), which we get by applying me result of the unification
process to the resolvent. The resolution process can then proceed
from there to discover whether mortal (Marcus) leads to a
contradiction with other available clauses.

Algorithm: Resolution in predicate logic

1. Convert all the statements of F to clause form.
2. Negate P and convert the result to clause form. Add it to the-set of clauses
obtained in 1.
3. Repeat until a contradiction is found, no progress can be made, or a
predetermined amount of effort has been expended.
(a) Select two clauses. Call these the parent clauses.
(b) Resolve them together. The resolvent will be the disjunction of all
the literals of both parent clauses with appropriate substitutions
performed and with the following exception: If there is one pair of
literals TI and ¬T2 such that one of the parent clauses contains T1
 AI 18

and the other contains ¬T2 and if T1 and T2 are unifiable, then
neither T1 nor T2 should appear in the resolvent. We call T1 and T2
Complementary Literals. Use the substitution produced by the
unification to create the resolvent. If there is more than one pair of
complementary literals, only one pair should be omitted from the
resolvent.
(c) If the resolvent is the empty clause, then a contradiction has been
found. If it is not, then add it to the set of c1auses available to the
procedure.
 There are systematic strategies for making the choice of clauses to
resolve together each step so that we will find a contradiction if one
exists. This can speed up the process considerably.
Only resolve pairs of clauses that contain complementary
literals, since only such resolutions produce new clauses that
are harder to satisfy than their parents. To facilitate this, index
clauses by the predicates they contain, combined with an
indication of whether the predicate is negated Then, given a
particular clause, possible resolvent that contain a
complementary occurrence of one of its predicates can be
 located directly.
Eliminate certain clauses as soon as they are generated so that
they cannot participate in later resolutions. Two kinds of
 clauses should be eliminated:
  Tautologies (which can never be unsatisfied) and
  Clauses that are subsumed by other clauses. For example, P⋁Q can be subsumed by P.
Whenever possible, resolve either with one of- the clauses that
is part of the statement we are trying to refute or with a clause
 generated by a resolution with such a clause. This is called the

“set-of-support strategy”.
Whenever possible, resolve with clauses that have a single
literal. Such resolutions generate new clauses with fewer literals
than the larger of their parent clauses and thus are probably
closer to the goal of a resolvent with zero terms. This method is
called the “unit-preference-strategy”.
 AI 19

Example 1:

Resolution Proof:
 AI 20

Example 2:
 AI 21

Resolution Proof:

3.13 Natural Deduction:

 We used resolution as an easily implementable proof procedure that
relies for its simplicity on a uniform representation of the statements
 it uses.
 Unfortunately, in uniformity, everything looks the same. Since
everything looks the same, there is no easy way to select those
statements that are the most likely to be useful in solving a particular
problem.
 In converting everything to clause form, we often lose valuable
heuristic information that is contained in the original representation
of the facts.
 For example, suppose the fact: All judges who are not crooked are
well-educated, can be represented as:
∀x : judge(x) ⋁ ¬crooked(x)  educated(x)
 AI 22

In this form, the statement suggests a way of deducing that someone

is educated. But when the same statement is converted to clause
form:
¬judge(x) V crooked(x) V educated(x)
It is a way of deducing that someone is not a judge by showing that he
is not crooked and not educated. Of course, in a logical sense, it is.
But it is almost certainly not the best way, or even a very good way, to
go about showing that someone is not a Judge. The heuristic
information contained in the original statement has been lost in the
transformation.
 Another problem with the use of resolution as the basis of a theorem-
proving system is that people do not think in resolution. Thus it is
very difficult for a person to interact with a resolution theorem prover,
either to give it advice or to be given advice by it. Since proving very
hard things is something that computers still do poorly, it is
important from a practical standpoint that such interaction be
 possible.
 Natural deduction is describes a melange of techniques used in
combination to solve problems that are not tractable by any one
 method alone.
 One common technique is to arrange knowledge, not by predicates,
but rather by the objects involved in the predicates.
 AI 23

Unit- III
Artificial Intelligence: Logic Concepts
(Open Elective –I)
Assignment-Cum-Tutorial Questions
Objective Questions
1. List standard logic symbols.
2. Represent “It is RAINING” in propositional logic.
3. Real-world facts written as logical propositions are called ______________.
4. Represent “Marcus was a man” in propositional and predicate logic.
5. In _________________ logic decision procedure does not exist. [ ]
(a) Propositional (b) Predicate (c) First order (d) none

6. _____________ and ___________ attributes play an important role in process

of reasoning.
7. ________________property uses the attributes isa and instance. [ ]
(a) Polymorphism (b) Inheritance (c) Reasoning (d) both b & c

8. The attribute instance is __________ predicate. [ ]

(a) Unary (b) Binary (c) Ternary (d) none

9. Resolution produces proofs by ___________________. [ ]

(a) Contradiction (b) Reasoning (c) Refutation (d) none

10. Define the term “Resolution”.

11. Moving all quantifiers to the left of the formula without changing their
relative order is called _____________________.
12. After resolving the below facts, we get_____________________.

winter V summer

¬winter V cold
13. The resultant clause after resolution is called______________. [ ]
(a) Implicant (b) Resolvent (c) fact (d) instance

14. The substitutions in Unification algorithm are applied from __. [ ]

(a) right to left (b) left to right (c) any order (d) none
 AI 24

15. Resolving clauses that have a single literal first is called _______________
strategy.
16. Resolving either with one of- the clauses that is part of the statement we
are trying to refute or with a clause generated by a resolution with such
a clause- This is called the ____________ strategy.

II) Descriptive Questions

1. Represent the following sentences in predicate logic:

1. Marcus was a man.
2. Marcus was a Pompeian.
3. All Pompeians were Romans.
4. Caesar was a ruler.

6. Everyone is loyal to someone.

7. People only try to assassinate rulers they are not loyal to.
8. Marcus tried to assassinate Caesar.

2. Answer the question: Was Marcus loyal to Caesar? using the above
facts.
3. Represent the following facts using isa and instance attributes:
1. Marcus was a man.
2. Marcus was a Pompeian.
3. All Pompeians were Romans.
4. Caesar was a ruler.
5. All Romans were either loyal to Caesar or hated him.
4. Enumerate the steps in the process of resolution.
5. Explain the steps of converting a fact to Clause form.
6. Convert the following fact into clause form:
All Romans who know Marcus either hate Caesar or think that anyone
who hates anyone is crazy.
7. Explain the algorithm for Propositional resolution.
8. Explain Unification Algorithm.
9. Explain the algorithm for resolution in predicate logic.
10. Determine the possible substitutions for unifying the following facts:
hate (x, y)
hate (Marcus, z)
11. Perform resolution in predicate logic:
 AI 25

12. Explain about Natural Deduction?

 AI Unit-4
UNIT - IV: Knowledge representation
Syllabus:
Introduction, approaches to knowledge representation, knowledge
representation using semantic network, extended semantic networks for KR,
knowledge representation using frames.
Advanced knowledge representation techniques: Introduction, conceptual
dependency theory, script structure, semantic web.

Outcomes:
Student will be able to:

Representations and Mappings


In order to solve the complex problems encountered in artificial
intelligence; we need both a large amount of knowledge and some

mechanisms for  manipulating that knowledge to create solutions to
new problems.

 of representing knowledge (facts) have been exploited in
A variety of ways
Al programs.
  We deal with two different kinds of entities:
 Facts: truths in some relevant world, these are the things we
 want to represent.
 Representations of facts in some chosen formalism. These are
the things we will actually be able to manipulate.

  These entities are at two levels:
 The knowledge level, at which facts (including each agent's
 behaviour. and current goals) are described.
 The symbol level, at which representations of objects at the
knowledge level are defined in terms of symbols that can be
manipulated by programs.
 AI Unit-4
  The two-way mappings exist between facts and representations called
“Representation mappings”. The forward representation mapping
maps from facts to representations. The backward representation
 mapping maps from representations to facts.

One representation of facts is: natural language (particularly
English) sentences. Regardless of the representation for facts that we
use in a program, we may also need to be concerned with an English
  in order to facilitate getting information
representation of those facts
into and out of the system.

In this case, we must have mapping function from English sentences
 to 
the representation we are actually going to use and back to
sentences.
 
Consider the English sentence:
Spot is a dog.
The fact represented by that English sentence can also be represented
in logic as:
dog (Spot)
Suppose that we also have a logical representation of the fact that all
dogs have tails: 
∀x: dog (x) hastail (x)

Then, using the deductive mechanisms of logic, we may generate the

new representation object:
hastail(Spot)
Using an appropriate backward mapping function, we could then
generate the English sentence:
Spot has a tail

The available mapping functions are not one-to-one; they are many .
to-many relations. In other words, each object in the domain may map
  elements in the domain
to several elements in the range, and several
may map to the same element of the range.)

 Consider an example of the  mappings involving English representations
of facts. The two sentences:
 "All dogs have tails" and "Every dog has a tail"
Both represent the same fact, namely that every dog has at least one
 tail.
On the other hand, the former could represent either the tact that
every dog has at least one tail or the fact that each dog has several
 tails.
The latter may represent either the fact that every dog has at least one
 tail or the fact that there is a tail that every dog has.

So, when we try to convert English sentences into some other 
representation, such as logical propositions, we must first decide what
 AI Unit-4
facts the sentences represent and then convert those facts into the
new representation.

Approaches to Knowledge Representation


 
A good system for the representation of knowledge in a particular domain
should possess the following four properties:
 Representational Adequacy: the ability to represent all of the
 kinds of knowledge that are needed in that domain.
 Inferential Adequacy: the ability to manipulate the
representational structures in such a way as to derive new
 structures corresponding to new knowledge inferred from old.
 Inferential efficiency: the ability to incorporate into the
knowledge structure additional information that can be used to
focus the attention of the inference mechanisms in the most
 promising directions.
 Acquisitional Efficiency: the ability to acquire new information
 easily.

But, no single system that optimizes all of the capabilities for all kinds
of knowledge has yet been found. As a result, multiple techniques for
knowledge representation exist. Many programs rely on more than one 
technique. Following are some knowledge representation techniques:
 AI Unit-4

Simple Relational Knowledge:


 is as a set of relations
The simplest way to represent declarative facts
of the same sort used in database systems.


The reason that this representation is simple is that standing alone
it provides very weak inferential capabilities, but knowledge

represented in thisform may serve as the input to more powerful
inference engines.

The relational knowledge corresponds to a set or attributes and
  that together describe the objects of the
associated values
knowledge base.

 knowledge is what database
Providing support for relational
systems are designed to do.

Inheritable Knowledge


 Knowledge about objects, their attributes, and their
values need not
be as simple as that shown in relational knowledge.

In particular, it is possible to augment the basic representation
with inference mechanisms that operate on the structure of the
 representation.

 For this to be effective, the structure must be 
designed to correspond
to the inference mechanisms that are desired.

One of the most useful forms of inference is property inheritance,
  and values
in which elements of specific classes inherit attributes
from more general classes in which they are included.

In order to support property inheritance, objects must be organized
  and classes must be arranged in a generalization
into classes
hierarchy.

Lines represent attributes. Boxed nodes represent objects and
values of attributes of objects. The arrows on the lines point from

an object to its value along the corresponding attribute line. The
structure called slot-and-filler structure.
 AI Unit-4


 Each individual frame represents the collection of attributes  and values
associated with a particular node. This structure is called a
Semantic network or a collection of Frames.

Baseball-Player isa: Adult-Male

bars: (EQUAL handed)
height: 6-1
batting-average: .252
Basic mechanism of inheritance:

Algorithm: Property Inheritance

To retrieve a value V, for attribute A of an instance object O:

1. Find O in the knowledge base.
2. If there is a value there for the attribute A, report that value.
3. Otherwise, see if there is a value for the attribute instance. If not,
then fail.
4. Otherwise, move to the node corresponding to that value and look
for a value for the attribute A. If one is found, report it.
5. Otherwise, do until there is no value for the isa attribute or until an
answer is found:
(a) Get the value of the isa attribute and move to that node.
(b) See if there is value for the attribute A. If there is, repori it.
 AI Unit-4

 We can apply this procedure to our example knowledge base to derive
answers to the following queries:
 team (Pee-Wee-Reese) = Brooklyn-fladgers. This attribute had a
 value stored explicitly in the knowledge base.
 batting-average ( Three- Finger-Brown) = .106. Since there is no
value for batting average stored explicitly for Three Finger Brown,
we follow the instance attribute to Pitcher and extract the value
 stored there.
 height (Pee-Wee-Reese) = 6-1. This represents another default
inference. Notice here that because we get to it first, the more
specific tact about the height of baseball players overrides a more
 general fact about the height of adult males.
 bats (Three- Finger-Brown) =Right. To get a value for the
attribute bats required going up the isa hierarchy to the class
Baseball-Player. But what we found there was not a value but a
rule for computing a value. This rule required another value as
input. So the entire process must be begun again recursively to
find a value for handed. This time, it is necessary to go all the way
up to Person to discover that the default value for handedness for
people is Right. Now the rule for bats can be applied, producing the
result Right.

Inferential Knowledge


 traditional logic is necessary to describe the inferences that
The power of
are needed.

This knowledge is useless unless there is also an inference procedure
 that can exploit it The required inference procedure is one that
implements the standard logical rules of inference.

The procedures, may reason forward from given facts to conclusions,
or may reason backward from desired conclusions to given facts. One

of the most commonly used of these procedures is Resolution, which
exploits a proof by contradiction strategy.

Procedural Knowledge
  This kind of knowledge is operational.
  Procedural knowledge specifies what to do and when to do.



Procedural knowledge can be represented in programs in many ways.
The most common way is simply as code (in some programming

language such as LISP) for doing something. The machine uses the
knowledge when it executes the code to perform a task.

 procedural knowledge gets low scores for
But, this way of representing
the following properties:
 AI Unit-4

 Inferential adequacy, because it is very difficult to write a

 program that can reason about another programs behaviour.
 Acquisitional efficiency, because the process of updating and
 debugging large pieces of code becomes unwieldy.


The most commonly used technique for representing procedural
knowledge in Al Programs is the use of production rules.

Semantic Nets

 
The main idea behind semantic nets is that the meaning of a concept
comes from the ways in which it is connected to other concepts.

In a semantic net, information is represented as a set of nodes
connected to each other by a set of labeled arcs, which represent
relationships among the nodes.


This network contains examples of both the isa and instance relations,
and domain-specific relations like team and uniform-color. In  this
 network, we could use inheritance to derive the additional relation:
 has-part (Pee- Wee-Reese, Nose)

Earlier, semantic nets were used to find relationships among objects
 by spreading activation out from each of two nodes and seeing where
the activation met. This process is called intersection search.

Using this process, it is possible to use the network to answer
questions such as "What is the connection between the Brooklyn
Dodgers and blue?" This kind of reasoning exploits one of the
important advantages that slot-and-filler structures have over purely
logical representations because it takes advantage of the entity-
based organization
 of knowledge that slot-and-filler representations
provide.
 AI Unit-4
Representing Non binary Predicates:



Semantic nets are a natural way to represent relationships that would
appear as ground instances of binary predicates in predicate logic.
  Some of the arcs can be represented in logic as:

  isa(Person, mammal)
  instance(Pee-Wee-Reese, Person)
  team( Pee-Wee-Reese, Brooklyn- Dodgers)
  uniform color(Pee-Wee-Reese, Blue)


Unary predicates in logic can be thought of as binary predicates using
 general-purpose predicates, such as isa and instance. For example:

Man (Marcus) could be rewritten as instance (Marcus, Man)



Three or more place predicates can also be converted to a binary form
by creating one new object representing the entire predicate statement
 and then introducing binary predicates to describe the relationship to
this new object of each of the original arguments.
  For example, score (Cubs, Dodger, 5-3)
This can be represented in a semantic net by creating a node to
represent the specific game and then relating each of the three pieces
of information to it.


The sentence: John gave  the book to Mary could be represented by
the network as follows:
 AI Unit-4

There should be a difference between a link that defines a new entity and
 one that relates two existing entities. Consider the net:






Both nodes represent objects that exist independently of their
 relationship to each other.

Suppose we want to represent the fact that: John is taller than Bill,
using the net:

The nodes HI and 112 are new concepts representing John's height
and Bill's height, respectively. They are defined by their relationships
to the nodes John and Hill.

Sometimes it is useful to introduce the arc value to make this

distinction clear. Thus we might use the following net to represent the
fact that John is 6 feet tall and that he is taller than Bill.
 AI Unit-4

Extended semantic networks for KR

  Partitioned Semantic Networks allow for:
• Propositions to be made without commitment to truth.
• Expressions to be quantified.

  consist of groups of
The basic idea is to break network into spaces which
nodes and arcs and regard each space as a node.


By partitioning the semantic net into a hierarchical set of spaces, each
partition corresponds to the scope of one or more variables.
  Consider the statement: The dog bit the mail carrier.
The nodes Dogs, Bite, and Mail-Carrier represent the classes of dogs,
bitings and mail carriers. Respectively, while the nodes d, b and m
represent a particular dog, a particular biting and a particular mail
carrier. This fact can easily be represented by a single net with no
partitioning.

  But now suppose that we want to represent the fact:

Every dog has bitten a mail carrier.
∀x: Dog(x)

∃y: Mail-Carrier(y) Λ Bite(x, y)

To represent this fact, it is necessary to encode the scope of the

universally quantified variable x. This can be done using partitioning
net.

Node g is an instance of the special class GS of general statements

about the world (i.e.. those with universal quantifiers). Every, element
 AI Unit-4
of GS has at least two attributes: a form, which states the relation
that is being asserted, and one or more ∀ connections, one for each of
the universally quantified variables. In this example, there is only one
such variable d, which can stand for any element of the class Dogs.
The other two variables in the form, b and m are existentially
quantified. In other words, for every dog d, there exists a biting event
b, and a mail carrier, m. such that d is the assailant of b and m is the
victim.
 
Consider the sentence:
Every dog in town has bitten the constable.
In this net, the node c representing the victim lies outside the form of
the general statement. Thus it is not viewed as an existentially
quantified variable whose value may depend on the value of d.

 
Consider the sentence:

Every dog has bitten every mail carrier.

In this case, g has two ∀ links: one pointing to d, which represents any dog, and one pointing to m, representing any mail carrier.

Space S is included in space SA. Whenever is search process operates

in a partitioned semantic net, it can explore nodes and arcs in the
space from which it starts and in other spaces that contain the
 AI Unit-4
starting point, but it cannot go downward, except in special
circumstances, such as when a form arc is being traversed. From
node d, it can be determined that d must be a dog. But if we were to
start, at the node Dogs and search for all known instances of dogs by
traversing isa links, we would not find d since it and the link to it are
in the space S1, which is at a lower level than space S4, which
contains Dogs. This is important, since d does not stand for a
particular dog; it is merely a variable that can be instantiated with a
value that represents a dog.
Frames

A frame is a collection of attributes (usually called slots) and
 
associated values (and possibly constraints on values) that describes
some entity in the world.

A single frame taken alone is rarely useful. Instead, we build frame
 other.
systems out of collection of frames that are connected to each
The value of an attribute of one frame may be another frame.

Frames as Sets and Instances:


Set theory provides a good basis for understanding frame systems.
 Each framerepresents either a class (a set) or an instance (an element
of a class).
 
Consider the following frame system:
 AI Unit-4


The frames Person, Adult-Male, ML-Baseball Player (corresponding to
major league baseball players), Pitcher, and ML-Baseball-Team (for major
 Brooklyn-Dodgers are instances.
league baseball team) are all classes. The frames Pee-Wee-Reese and


The set of adult males is a subset of the set of people. The set of major
league baseball players is a subset of the set of adult males, and so forth.
The instance relation corresponds to the relation “element-of”. Pee Wee
Reese is an element of the set of fielders. Thus he is also an element of all
 people.
of the supersets of fielders, including major league baseball players and


A  are two kinds of attributes that can be
class represents a set; there
associated with it. They are:
  attributes about the set itself, and
 attributes that are to be inherited-by each element of the set
We indicate the difference between these two by prefixing the latter with
an asterisk (*).For example, consider the class ML-Basebalf-Player.

It is important to distinguish between regular classes, whose elements
are individual entities, and metaclasses, which are special classes whose
elements are themselves classes. A class is now an element of (instance)
some class (or classes) as well as a subclass (isa) of one or more classes.
A class inherits properties from the class of which it is an instance, just
as any instance does. In addition, a class  passes inheritable properties
down from its super classes to its instances.
 AI Unit-4


The most basic metaclass is the class Class. It represents the set of
all classes. All classes are instances of it, either directly or through
one of its subclasses. In the example, Team is a subclass (subset) of
Class and ML-Baseball-Team is a subclass of Team. The class Class
introduces the attribute cardinality,  which is to be inherited by all
instances of Class (including itself).

  Ways in which classes are related to each other:

  Class1 can be a subset of Class2.
  If Class2 is a metaclass, then Class1 can be an instance of Class2.
 Mutually-disjoint-with: which relates a class to one or more other
classes that are guaranteed to have no elements in common with
 it.
 Is-covered-by: which relates a class to a set of subclasses, the
union of which is equal to it. If a class is covered-by a set of
 mutually disjoint classes, then S is called a partition of the class.

 
The following are the properties the frames would be able to represent
and use in reasoning:
  The classes to which the attribute can be attached
  Constraints on either the type or the value of the attribute.
 A value that all instances of a class must have by the definition of
 the class.
  A default value for the attribute.
 Rules for inheriting values for the attribute. The usual rule is to
inherit down isa and instance links.
 AI Unit-4

 Rules for computing a value separately from inheritance.

Conceptual Dependency

Semantic networks and frame systems may have specialized links and
inference procedures, but there are no hard and fast rules about what

kinds of objects and links arc good in general for knowledge
representation.

 ConceptualDependency specifies what types of objects and relations are
permitted.

Conceptual Dependency (CD) is a theory of how to represent the kind
of knowledge about events that is usually contained in natural
  sentences. The goal is to represent the knowledge in a way
language
that:
  Facilitates drawing inferences from the sentences.
 Is independent of the language in which the sentences were
 originally stated.

The CD representation of a sentence is built not out of primitives
corresponding to the words used in the sentence, but rather out of
conceptual primitives that can becombined to form the meanings of
 words in any particular language.

Semantic nets provide only a structure into which nodes representing
information at any level can be placed. Conceptual dependency
provides both a structure and a specific set of primitives, at a
particular level of granularity out of whichrepresentations of
 particular pieces information can be constructed.


Consider a Simple Conceptual Dependency Representation of : I gave
the man a book, where the symbols have the following meanings:

 
Arrows indicate direction of dependency.
 
Double arrow indicates two way links between actor and action.
 
p indicates past tense.

ATRANS is one of the primitive acts used by the theory. It indicates
 transfer of possession.
  O indicates the object case relation.
 R indicates the recipient case relation.
 
In CD, representations of actions are built from a set of primitive acts:
 AI Unit-4


The set of CD building blocks is the set of allowable dependencies
among the conceptualizations described in a sentence. There are tour

primitive conceptual categories from which dependency structures
can be built. These are:
 ACTS Actions
 PPs Objects (picture producers)
 AAs Modifiers of actions (action aiders)
 PAs Modifiers of PPs (picture aiders)
  
The Dependencies of CD:
 Rule 1 describes the relationship between an actor and the event he
or she causes. This is a two-way dependency since neither actor
nor event can be considered primary. The letter p above the
 dependency link indicates past tense.
 Rule 2 describes the relationship between a PP and a PA that is
 being asserted to describe it.
 Rule 3 describes the relationship between two PPs, one of which
 belongs to the set defined by the other.
 Rule 4 describes the relationship a PP and an attribute that has
already been predicated of it. The direction of the arrow is toward
 the PP being described.
 Rule 5 describes the relationship between two PPs, one of which
provides a particular kind of information about the other. The three
most common types of information to be provided in this way are:
  
possession (shown as POSSBY)
  
location (shown as LOC), and
  
physical containment (shown as CONT)
 
The direction of is again toward the concept being described
 AI Unit-4

 Rule 6 describes the relationship between an ACT and the PP that is

the object oh that ACT. The direction of the arrow is toward the
ACT since the context of the specific ACT determines the meaning
 of the object relation.
 Rule 7 describes the relationship between an ACT and the source
 and the recipient of the ACT.
 Rule 8 describes the relationship between an ACT and the
instrument with which it is performed. The instrument must
 always be a full conceptualization, not just a single physical object.
 Rule 9 describes the relationship between an ACT and its physical
 source and destination.
 Rule 10 represents the relationship between a PP and a state in
 which it started and another in which it ended.
 Rule 10 describes the relationship between one conceptualization
and another that causes it. The arrows indicate dependency of one
conceptualization on another and point in the opposite direction of
the implication arrows. The two forms of the rule describe the
 cause of an action and the cause of a state change.
 Rule 12 describes the relationship between a conceptualization and
 the time at which the event it describes occurred.
 Rule 13 describes the relation between one conceptualization and
 another that is the time of the first.
 Rule 14 describes the relationship between a conceptualization and
the place at which it occurred.
 
The set of conceptual tenses in a CD are:
 p Past
 f Future
 t Transition
  txStart transition
  tyFinished transition
  kContinuing
 ?Interrogative
 nil Present
  delta Timeless
 cConditional
 AI Unit-4

 Consider the sentence: Since smoking can kill you, I stopped.

 AI Unit-4

Scripts

CD is a mechanism for representing and reasoning about events. A
script is a structure 
that describes a stereotyped sequence of events in
 a particular context.

 Script is a mechanismfor representing knowledge about common
sequences of events.

A script consists of a set of slots. Associated with each slot may be
 some information about what kinds of values it may contain as well as
a default value to be used if no other information is available.
 
Consider the following example of a restaurant script:

 
The important components of a script are:
 AI Unit-4

 Entry conditions: Conditions that must, be satisfied before the

 events described in the script can occur.
 Result: Conditions that will, be true after the events described in
 the script have occurred.
 Props: Slots representing objects that are involved in the events
described in the script. The presence of these objects can be
 inferred even if they are not mentioned explicitly.
 Role: Slots representing people who are involved in the events
 described in the script.
 Track: The specific variation on a more general pattern that is
represented by this particular script. Different tracks of the same
 script will share many but not all components.
 Scenes: The actual sequences of events that occur.


Scripts are useful because, in the real world, there are patterns to the
 occurrence of events. These  patterns arise because of causal
relationships between events.

Agents will perform one action so that they will then be able to
 performanother. The events described in a script form a giant causal
chain.

The beginning of the chain is the set of entry conditions which enable
the first events of the script to occur. The end of the chain is the set of
results which may enable later events or event sequences to occur.

Within the chain, events are connected both to earlier events that
make them possible and to later events that they enable.

  to indicate how events that were mentioned
Scripts can also be used
relate to each other.

  be activated. There are
Before a particular script can be applied, it must
two ways in which a script can be activated:
 For fleeting scripts (ones that are mentioned briefly and may he
referred to again but are not central to the situation), it may be
sufficient to store a pointer to the script so that it can be accessed
 later if necessary.
 For non-fleeting scripts it is appropriate to activate the script fully
and to attempt to fill in its slots with particular objects and people
 involved in the current situation.

Once a script has been activated, there are varieties of ways in which
it can be useful in interpreting a particular situation. The most
important of these is the 
ability to predict events that have not
 explicitly been observed.
For example:
 AI Unit-4
John went out to a restaurant last night. He ordered steak.
When he paid for it, he noticed that he was running out of
money. He hurried home since it had started to rain.
If you were then asked the question:
Did John eat dinner last night?
By using the restaurant script, a computer question-answerer would
also be able to infer that John ate dinner, since the restaurant script
could have been activated. Since all of the events in the story
correspond to the sequence of events predicted by the script, the
program could infer that the entire sequence predicted by the script
occurred normally. Thus it could conclude that John ate.

 
Scripts provide a way of building a single coherent interpretation
from a collection of observations. Consider, for example:
Susan went out to lunch. Site sat down at table and called the
waitress. The waitress brought her a menu and she ordered a
 hamburger.
 Now consider the question:
 Why did the waitress bring Susan a menu?
 The script provides two possible answers to that question:
  Because Susan asked her to.
 So that Susan could decide what she wanted to eat.

Assignment-Cum-Tutorial Questions
I) Objective Questions
1. The two different kinds of entities in AI are _____________ and ______________.
2. The __________ mappings exist between facts and representations.
(a) One-way (b) two-way (c) no mapping (d) both (a) & (b) [ ]

3. The forward representation mapping maps from _____________ to

__________________.
(a) representations, facts (b) facts, representations
(c) facts, facts (d) representations, representations [ ]

4. The ability to represent all of the kinds of knowledge that are needed in that
domain is called ______________.

(a) Referential Efficiency (b) Representational Adequacy

(c) Inferential Efficiency (d) Acquisitional Efficiency [ ]
5. The ability to acquire new information easily is called ___________. [ ]

(a) Referential Efficiency (b) Representational Adequacy

 AI Unit-4
(c) Inferential Efficiency (d) Acquisitional Efficiency
6. The two important attributes of inheritance are ________ and _________.

7. Weak slot-and-filler structures are ___________and ______________.

8. Strong slot-and-filler structures are ___________and ______________.

9. Procedural knowledge is ______________.

(a) Declarative (b) Operational (c) progressive (d) none [ ]

10. Procedural Knowledge get low scores for the properties______

(a) Inferential Adequacy (b) Acquisitional Efficiency

(c) Inferential Efficiency (d) both (a) and (b) [ ]

11. The most commonly used technique for representing procedural knowledge in
Al Programs is the use of _____________.

(a) Production rules (b) symbols (c) facts (d) both (b) & (c) [ ]

12. The structure in which information is represented as a set of nodes connected

to each other by a set of labeled arcs, which represent relationships among the
nodes is called _____________.

13. Define the term “frame”.

14. ___________ theory provides a good basis for understanding frame systems.

(a) set (b) Graphics (c) Logic (d) none [ ]

15. The classes whose elements are themselves classes are called _______.
(a) Sub class (b) Base class (c) Meta class (d) Parent class [ ]

16. ______________ is a theory of how to represent the kind of knowledge about

events that is usually contained in natural language sentences.
17. The primitive that represents transfer of an abstract relationship is ________.
(a) PTRANS (b) ATRANS (c) MOVE (d) GRASP [ ]

18. A _______________ is a structure that describes a stereotyped sequence of events

in a particular context.

II) Descriptive Questions

1. Enlist the four properties that a knowledge representation system must

have?
2. Explain four knowledge representation techniques.
3. Enumerate the basic mechanism of retrieving a value of an attribute, using
inheritance.
 AI Unit-4
4. How non binary predicates are represented using semantic net. Explain with
suitable example.
5. Represent the following facts using semantic nets:
John gave the book to Mary
John is 6 feet tall and that he is taller than Bill.
6. Represent the following facts using partitioned semantic nets:
The dog bit the mail carrier.
Every dog has bitten a mail carrier.
Every dog in town has bitten the constable.
Every dog has bitten every mail carrier.

7. Justify the statement- “Set theory provides a good basis for understanding
frame systems”.
8. List the ways in which classes are related to each other in frames, with
suitable example?
9. List the set of primitives and conceptual tenses used in Conceptual
Dependency.
10. Explain the rules used in Conceptual Dependency.

11. Represent the following sentence in CD:

Since smoking can kill you, I stopped.

12. Describe the important components of a script, with a suitable example.
 AI Unit-5
Unit 5
Reasoning in Uncertain Situations

1. Non-Monotonic Reasoning

A non-monotonic logic is a formal logic whose consequence relation is

not monotonic. In other words, non-monotonic logics are devised to capture and
represent de feasible inferences (cf. de feasible reasoning), i.e., a kind of inference in
which reasoners draws tentative conclusions, enabling reasoners to retract their
[1]
conclusion(s) based on further evidence. Most studied formal logics have a
monotonic consequence relation, meaning that adding a formula to a theory never
produces a reduction of its set of consequences. Intuitively, monotonicity indicates
that learning a new piece of knowledge cannot reduce the set of what is known. A
monotonic logic cannot handle various reasoning tasks such as reasoning by default
(consequences may be derived only because of lack of evidence of the contrary),
abductive reasoning (consequences are only deduced as most likely explanations)

Rules with exceptions

Most rules we use in commonsense reasoning — like university professors teach,

birds fly, kids like ice-cream, Japanese cars are reliable — have exceptions. The rules
describe what the case is normally, but they do not necessarily hold without
exception. This is obviously in contrast with universally quantified formulas in first
order logic. The sentence
∀x (prof (x) ⊃ teaches (x))

Simply excludes the possibility of non-teaching university professors and thus cannot
be used to represent rules with exceptions. Of course, we can refine the sentence to
∀x ((prof (x) ∧ ¬abnormal (x)) ⊃ teaches (x)).
 AI Unit-5
However, to apply this rule, say to Professor Jones, we need to know whether
Professor Jones is exceptional (for instance, professors who are department Chairs do
not teach). Even if we assume that the unary predicate abnormal (.) can be defined
precisely, which is rarely the case in practice as the list of possible exceptions is hard
if not impossible?
To complete, we will most often lack information to derive t hat Professor
Jones is not exceptional. We want to apply the rule even if all we know about Dr.
Jones is that she is a professor at a university. If we later learn she is a department
Chair — well, then we have to retract our former conclusion about her teaching
classes. Such scenarios can only be handled with a non-monotonic reasoning
formalism.

2. Truth Maintenance Systems

A truth maintenance system, or TMS, is a knowledge representation

method for representing both beliefs and their dependencies and an
algorithm called the "truth maintenance algorithm" that manipulates and
maintains the dependencies. The name truth maintenance is due to the
ability of these systems to restore consistency.

A truth maintenance system maintains consistency between old believed

knowledge and current believed knowledge in the knowledge base (KB)
through revision. If the current believed statements contradict the
knowledge in the KB, then the KB is updated with the new knowledge. It
may happen that the same data will again be believed, and the previous
knowledge will be required in the KB. If the previous data are not present,
but may be required for new inference. But if the previous knowledge was in
the KB, then no retracing of the same knowledge is needed. The use of TMS
avoids such retracing; it keeps track of the contradictory data with the help
of a dependency record. This record reflects the retractions and additions
which makes the inference engine (IE) aware of its current belief set.
 AI Unit-5

Each statement having at least one valid justification is made a part of the
current belief set. When a contradiction is found, the statement(s)
responsible for the contradiction are identified and the records are
appropriately updated. This process is called dependency-directed
backtracking.

The TMS algorithm maintains the records in the form of a dependency

network. Each node in the network is an entry in the KB (a premise,
antecedent, or inference rule etc.) Each arc of the network represent the
inference steps through which the node was derived.

A premise is a fundamental belief which is assumed to be true. They do not

need justifications. The set of premises are the basis from which
justifications for all other nodes will be derived.

There are two types of justification for a node. They are:

1. Support List [SL]

2. Conditional Proof (CP)
Many kinds of truth maintenance systems exist. Two major types are single-
context and multi-context truth maintenance. In single context systems,
consistency is maintained among all facts in memory (KB) and relates to the
notion of consistency found in classical logic. Multi-context systems support
paraconsistency by allowing consistency to be relevant to a subset of facts in
memory, a context, according to the history of logical inference.

Inference Truth
Engine Maintenance
system

TMS GOALS
 AI Unit-5
1.Provide justification for conclusions
2.Recognise inconsistencies
3support default resoning
4.Remember derivations computed previously
5support dependency driven back tracking

Three categories of TMS:

 Justification-Based TMS (JTMS)


1. One element of a TMS design space
2. Simplest
3. Good model for most dependency systems
4. Can quickly focus on how to use it

JTMS nodes
 Each belief represented by a TMS node

 Nodes are associated 1:1 with assertions

 The label of a node represents the belief

 Status of the corresponding problem solv

 AI Unit-5
Fact
 The relationships between beliefs are
Expressed by the justifications it participates
In

JTMS Labels:

JTMS Justifications:

Must be Horn clauses

 Horn clause

Nomenclature

 Consequent is the node whose belief
is supported by the justification

 Antecedents are the beliefs which, when IN, support the consequent

 Informant records information from external systems

Enforcing constraints between beliefs

1 A node is IN when either:

 it is an enabled assumption or promise
 AI Unit-5

 there exists a justification for it whose antecedents are all IN

2 Assumptions underlying a belief can be found by backchaining through
supporting justifications
3. JTMS operations must preserve wellfounded support

Assumption-Based TMS (ATMS)

ATMS is a general search - control mechanism, that can be coupled with

domain-specific problem to solve a wide range of problem

Start with a JTMS

Label become more complex than: in, out

ATMS data structures

 Assumption

 Environments

 Labels

Justifications:

Environments = set of assumption

Label = set of environment
Type of labels:
Sound - All assumptions in any environment within the label of a
node being true is a sufficient condition to derive that node:
 AI Unit-5

Consistent - No environment in the label of a node, other than the nogood

node, describes an impossible world:

Minimal - The label does not contain possible worlds that are less
general than one of the other possible worlds it contains (i.e.
environments that are supersets of other environments in the label):
Complete - The label of each node, other than the nogood node, describes
all possible worlds in which that node can be inferred:

Logic-Based TMS (LTMS)

Logic Truth Maintenance Systems
 ideal for real time propositional reasoning

LTMS database summatory

 contains clauses describing component behavior in each mode

3. Logics for Non-Monotonic

Reasoning Default Reasoning:

1. Nonmonotonic Logic .

Example: let the KB contain:

Typically birds fly.
Penguins do not fly.
Tweety is a bird.

We cannot represent a rule such as "typically birds fly" as

∀ x (bird ( x ) ∧ ¬exception ( x ) → fly ( x)) and then to add
∀ x (exception ( x ) ↔ penguin ( x ) ∨ ostrich ( x ) ∨canary ( x ) ∨. . . )
 AI Unit-5

2. Default Logic

Default reasoning is common. It appears when we apply the Closed-

World Assumption to derive negative information, when we use
inference rules that admit exceptions (rules that hold under the
normality assumption), and when we use frame axioms to reason
about effects of actions. Ray Reiter, who provided one of the most
robust formalizations of default reasoning, argued that understanding
default reasoning is of foremost importance for knowledge
representation and reasoning. According to Reiter defaults are meta-
rules of the form “in the absence of any information to the contrary ,
assume ...” and default reasoning consists of applying them .

Usual inference rules sanction the derivation of a formula whenever

some other formu-las are derived. In contrast, Reiter's defaults require
an additional consistency condition to hold. For instance, a default
rule normally, a university professor teaches is represented in Reiter's
default notation as

prof (x) : teaches (x) .

teaches (x)

It states that if prof (J ) is given or derived for a particular ground

term J (which may represent Prof. Jones, for instance) and teaches(J )
is consistent (there is no information that ¬teaches(J ) holds), then
teaches(J ) can be derived “by default”. The key question of course is:
consistent with what? Intuitively, teaches(J ) has to be consistent with
the whole set of formulas which one can “reasonably” accept based on
the available in-formation. Reiter's far-reaching contribution is that he
made this intuition formal. In his approach, depending on the choice
of applied defaults, different sets of formulas may be taken as
 AI Unit-5
providing context for deciding consistency. Reiter calls these different
sets extensions.
One can use extensions to define a skeptical inference relation (a
formula is skeptically entailed by a default theory if it belongs to all of
its extensions), or a credulous inference relation (a formula is
credulously entailed by a default theory if it belongs to at least one of
its extensions). In many applications such as diagnosis, planning and,
more generally in all the situations where defaults model constraints,
the extensions themselves are of interest as they represent different
solutions to a problem

4.Classical planning problem: Goal stack

What is planning in AI?

 The planning in Artificial Intelligence is about the decision making tasks

performed by the robots or computer programs to achieve a specific goal.

 The execution of planning is about choosing a sequence of actions with a
high likelihood to complete the specific task.

Blocks-World planning problem

 The blocks-world problem is known as Sussman Anomaly.

 Noninterleaved planners of the early 1970s were unable to solve this
problem, hence it is considered as anomalous.

 When two subgoals G1 and G2 are given, a noninterleaved planner
produces either a plan for G1 concatenated with a plan for G2, or vice-
versa.

 In blocks-world problem, three blocks labeled as 'A', 'B', 'C' are allowed to
rest on the flat surface. The given condition is that only one block can be
moved at a time to achieve the goal.

 The start state and goal state are shown in the following diagram.
 AI Unit-5

Goal stack planning

This is one of the most important planning algorithms, which is specifically
used by STRIPS.

 The stack is used in an algorithm to hold the action and satisfy the goal. A
knowledge base is used to hold the current state, actions.

 Goal stack is similar to a node in a search tree, where the branches are
created if there is a choice of an action.

The important steps of the algorithm are as stated below:


i. Start by pushing the original goal on the stack. Repeat this until the
stack becomes empty. If stack top is a compound goal, then push its
unsatisfied subgoals on the stack.
ii. If stack top is a single unsatisfied goal then, replace it by an action and
push the action’s precondition on the stack to satisfy the condition.
iii. If stack top is an action, pop it from the stack, execute it and change
the knowledge base by the effects of the action.
iv. If stack top is a satisfied goal, pop it from the stack.

Goal stack planning 1

(I do pushing the subgoals at the same step as the compound goal.)
order of subgoals is up to us; we could The have put On(B, C) on top
On(A, B)
On(B,C)
 AI Unit-5
OnTable(C)
On(A, B) ∧ On(B, C) ∧ OnTable(C)
KB={On(C,A),OnTable(A),OnTable(B),Clear(B),Clear(C)} plan = [ ]

The top of the stack is a single unsatisfied goal. We push the action
which
would achieve it, and its preconditions.
Goal stack planning 2

On(A, Table)
Clear(B)
Clear(A)
On(A, Table) ∧ Clear(A) ∧ Clear(B)

MOVE(A,x, B)
On(A, B)
On(B,C)
OnTable(C)
On(A, B) ∧ On(B, C) ∧ OnTable(C)
KB={On(C,A),OnTable(A),OnTable(B),Clear(B),Clear(C)} plan = [ ]

The top of the stack is a satisfied goal. We pop the stack (twice).

Goal stack planning 3:

Clear(A)
On(A, Table) ∧ Clear(A) ∧ Clear(B)

MOVE(A,Table, B)
On(A, B)
On(B,C)
OnTable(C)
On(A, B) ∧ On(B, C) ∧ OnTable(C)

KB={On(C,A),OnTable(A),OnTable(B),Clear(B),Clear(C)}
plan = [ ]
 AI Unit-5

The top of the stack is an unsatisfied goal. We push the action which
would achieve it, and its preconditions.

5. Hierarchical planning:
Hierarchy of actions

a. In terms of major action or minor action

b. Lower level activities would detail more precise steps for

accomplishing the higher level tasks.

Example:

Planning for ”Going to Goa this Cristmas”

c. Major Steps :

i. Hotel Booking

ii. Ticket Booking

iii. Reaching Goa

iv. Staying and enjoying there

v. Coming Back

d. Minor Steps :

i. Take a taxi to reach station / airport

ii. Have candle light dinner on beach

iii. Take photos

Motivation
 
Reduces the size of search space
 AI Unit-5

Instead of having to try out a large number of

possible plan ordering, plan hierarchies limit the ways in which
an agent can select and order its primitive operators

Example:
 
 180 overs : 15 spells (12 overs each)
 
 5 bowlers : 3 categories (2 pacer/2 spinner/1 pacer&1 spinner)
 
 Top level possibilities : 315
 
Total possibilities < 3*315 (much less than 5180)

If entire plan has to be synthesized at the level of most detailed actions, it

would be impossibly long.

Natural to 'intelligent' agent


 attempts to solve mere details, until major steps are in
Postpone
place.


Higher level plan may run into difficulties at a lower level, causing the
need to return
 to higher level again to produce appropriately ordered
sequence.

Planner:
 
 Identify a hierarchy of conditions
 
 Construct a plan in levels, postponing details to the next level
 
 Patch higher levels as details become visible
 
Demonstrated using ABSTRIPS

ABSTRIPS:
 
 Abstraction-Based STRIPS
 
Modified version of STRIPS that incorporates hierarchical planning
 AI Unit-5
Hierarchy in ABSTRIPS:


  reflect the intrinsic difficulty of achieving
Hierarchy of conditions
various conditions.
 
Indicated by criticality value.

Criticality:


  achievable, i.e.,
A operation having minimum criticality can be trivially
the operations having very less or no precondition.
 
Example : Opening makemytrip.com


 many preconditions to satisfy will
Similarly operation having
have higher criticality.

Patching in ABSTRIPS:



Each level starts with the goal stack that includes the plan obtained
in the higher levels.
 
The last item in the goal stack being the main goal.
 Artificial Intelligence Unit-6

UNIT - VI: Expert system and applications

Syllabus:
Introduction phases in building expert systems, expert system versus
traditional systems, rule-based expert systems blackboard systems truth
maintenance systems, application of expert systems, list of shells and tools.
Outcomes:
Student will be able to:
Introduction: Expert System


The expert systems are the computer applications developed to solve
complex problems in a particular domain, at the level of extra-ordinary
human intelligence and expertise.
 
 Characteristics of Expert Systems:
 High performance

 Understandable

 Reliable

 Highly responsive
 
 Capabilities of Expert Systems: The expert systems are capable of −
 Advising

 Instructing and assisting human in decision making

 Demonstrating

 Deriving a solution

 Diagnosis

 Explaining

 Interpreting input

 Predicting results

 Justifying the conclusion

 Suggesting alternative options to a problem
 
 They are incapable of :
 Substituting human decision makers

 Possessing human capabilities

 Producing accurate output for inadequate knowledge base
 Artificial Intelligence Unit-6

 Refining their own knowledge

Components of Expert Systems

 
 The components of ES include −
 Knowledge Base

 Inference Engine

 User Interface

Knowledge Base:
 
 It contains domain-specific and high-quality knowledge.

Knowledge is required to exhibit intelligence. The success of any ES
majorly depends
 upon the collection of highly accurate and precise
 knowledge.

What is Knowledge?-The data is collection of facts. The information is
organized as data and facts about the task domain. Data,
information, and past experience combined together are termed as

 knowledge.

Components of Knowledge Base: The knowledge  base of an ES is a
 store of both, factual and heuristic knowledge.
 Factual Knowledge − It is the information widely accepted by the
Knowledge Engineers and scholars in the task domain.
 Artificial Intelligence Unit-6

 Heuristic Knowledge − It is about practice, accurate judgement,

one’s ability of evaluation, and guessing.


Knowledge representation: It is the method used to organize and
formalize the knowledge
 in the knowledge base. It is in the form of IF-
 THEN-ELSE rules.

Knowledge Acquisition: The success of any expert system majorly
depends on the quality, completeness,  and accuracy of the
 information stored in the knowledge base.

The knowledge base is formed by readings from various experts,
scholars, and the Knowledge Engineers. The knowledge engineer is a
person with the qualities of empathy, quick learning, and case

 analyzing skills.

He acquires information from subject expert by recording,
interviewing, and observing him at work, etc. He then categorizes and
organizes the information in a meaningful way, in the form of IF-
THEN-ELSE rules, to be used by interference machine. The

knowledge engineer also monitors the development of the ES.

Inference Engine:

 Inference Engine is
Use of efficient procedures and rules by the
 essential in deducting a correct solution.

In case of knowledge-based ES, the Inference Engine acquires and
manipulates the knowledge
 from the knowledge base to arrive at a
particular solution.
 
 In case of rule based ES, it −
 Applies rules repeatedly to the facts, which are obtained from
earlier rule application.

 Adds new knowledge into the knowledge base if required.

 Resolves rules conflict when multiple rules are applicable to a
particular case.


To recommend a solution, the Inference Engine uses the following
 strategies −
 Forward Chaining
 Artificial Intelligence Unit-6

 Backward Chaining


Forward Chaining: It is a strategy of 
an expert system to answer the
 question, “What can happen next?”
 Here, the Inference Engine follows the chain of conditions and
derivations and finally deduces the outcome. It considers all the
facts and rules, and sorts them before concluding to a solution.

 This strategy is followed for working on conclusion, result, or
effect. For example, prediction of share market status as an effect
of changes in interest rates.


 system finds out the
Backward Chaining: With this strategy, an expert
 answer to the question, “Why this happened?”
 On the basis of what has already happened, the Inference Engine
tries to find out which conditions could have happened in the past for
this result. This strategy is followed for finding out cause or reason.
For example, diagnosis of blood cancer in humans.
 Artificial Intelligence Unit-6

User Interface:


User interface provides interaction between user of the ES and the ES
itself. It is generally Natural Language Processing so as to be used by
the user who is well-versed in the task domain. The user of the ES

need not be necessarily an expert in Artificial Intelligence.
 
 It explains how the ES has arrived at a particular recommendation.
 The explanation may appear in the following forms −
 Natural language displayed on screen.

 Verbal narrations in natural language.

 Listing of rule numbers displayed on the screen.


 
The user interface makes it easy to trace the credibility of the
deductions.
 
 Requirements of Efficient ES User Interface:
 It should help users to accomplish their goals in shortest possible
way.

 It should be designed to work for user’s existing or desired work
practices.

 Its technology should be adaptable to user’s requirements; not the
other way round.

 It should make efficient use of user input.

Expert Systems Limitations


No technology can offer easy and complete solution. Large systems
are costly, require significant development time, and computer
 resources. ESs have their limitations which include −
 Limitations of the technology

 Difficult knowledge acquisition

 ES are difficult to maintain

 High development costs

Applications of Expert System

 Artificial Intelligence Unit-6

 The following table shows where ES can be applied.

Application Description

Design Domain Camera lens design, automobile design.

Diagnosis Systems to deduce cause of disease

Medical Domain from observed data, conduction medical
operations on humans.

Comparing data continuously with observed

Monitoring Systems system or with prescribed behavior such as
leakage monitoring in long petroleum pipeline.

Controlling a physical process based on

Process Control Systems
monitoring.

Knowledge Domain Finding out faults in vehicles, computers.

Detection of possible fraud, suspicious

Finance/Commerce transactions, stock market trading, Airline
scheduling, cargo scheduling.

Expert System Technology


There are several levelsof ES technologies available. Expert systems
 technologies include −
 Expert System Development Environment − The ES
development environment includes hardware and tools. They are −

o Workstations, minicomputers, mainframes.
 Artificial Intelligence Unit-6

o High level Symbolic Programming Languages such as LISt

Programming (LISP) and PROgrammation en LOGique
(PROLOG).

o Large databases.

 Tools − They reduce the effort and cost involved in developing an

expert system to large extent.

o Powerful editors and debugging tools with multi-windows.

o They provide rapid prototyping

o Have Inbuilt definitions of model, knowledge representation,

and inference design.

 Shells − A shell is nothing but an expert system without

knowledge base. A shell provides the developers with knowledge
acquisition, inference engine, user interface, and explanation
facility. For example, few shells are given below −

o Java Expert System Shell (JESS) that provides fully developed
Java API for creating an expert system.

o Vidwan, a shell developed at the National Centre for Software
Technology, Mumbai in 1993. It enables knowledge encoding in
the form of IF-THEN rules.

Benefits of Expert Systems


 − They are easily available due to mass production of
Availability
 software.

Less Production Cost − Production cost is reasonable. This makes
 them affordable.

 Speed − They offer great speed. They reduce the amount of work an
individual puts in.
 
 Less Error Rate − Error rate is low as compared to human errors.

ReducingRisk − They can work in the environment dangerous to
humans.
 Artificial Intelligence Unit-6


Steady response − They work steadily without getting motional,
tensed or fatigued.

Phases in building Expert Systems

The process of ES development is iterative. Steps in developing the ES
include −
 
 Identify Problem Domain
 The problem must be suitable for an expert system to solve it.

 Find the experts in task domain for the ES project.

 Establish cost-effectiveness of the system.
 
 Design the System
 Identify the ES Technology

 Know and establish the degree of integration with the other
systems and databases.

 Realize how the concepts can represent the domain knowledge
best.
 
 Develop the Prototype
 From Knowledge Base: The knowledge engineer works to −
o Acquire domain knowledge from the expert.

o Represent it in the form of If-THEN-ELSE rules.
 
 Test and Refine the Prototype
 The knowledge engineer uses sample cases to test the prototype for
any deficiencies in performance.

 End users test the prototypes of the ES.
 
 Develop and Complete the ES
 Test and ensure the interaction of the ES with all elements of its
environment, including end users, databases, and other
information systems.

 Document the ES project well.

 Train the user to use ES.
 
Maintain the System
 Keep the knowledge base up-to-date by regular review and update.
 Artificial Intelligence Unit-6

 Cater for new interfaces with other information systems, as those

systems evolve.

Expert system versus traditional systems

Expert System Traditional System

The entire problem related expertise Problem expertise is encoded in both
is encoded in data structures only, program and data structures.
none is in programs.
The use of knowledge is vital. Data is used more efficiently than
knowledge.
These are capable of explaining how These are not capable of explaining a
a particular conclusion is reached particular conclusion for a problem.
and why requested information is These systems try to solve in a
needed during a process. straight forward manner.
Problems are solved more efficiently Not so efficient as an expert system
It uses the symbolic representations These are unable to express in
for knowledge i.e. the rules, different symbols. They just simplify the
forms of networks, frames, scripts problems in a straight forward
etc. and performs their inference manner and are incapable to express
through symbolic computations the “how, why” questions.
Problem solving tools those are No problem solving tools in specific.
present in expert system
Solution of the problem is more Solution of the problem may not be
accurate. more accurate.

Provide a clear separation of Do not separate knowledge from the

knowledge from its processing. control structure to process this
knowledge.
Process knowledge expressed in the Process data and use algorithms, a
form of rules and use symbolic series of well-defined operations, to
reasoning to solve problems in a solve general numerical problems.
narrow domain.
Trace the rules fired during a Do not explain how a particular
problem-solving session and explain result was obtained and why input
how a particular conclusion was data was needed.
reached and why specific data was
needed.
Permit inexact reasoning and can W ork only on problems where data
deal with incomplete, uncertain and is complete and exact.
fuzzy data.
Enhance the quality of problem Enhance the quality of problem
solving by adding new rules or solving by changing the program
adjusting old ones in the knowledge code, which affects both the
base. W hen new knowledge is knowledge and its processing,
acquired, changes are easy to making changes difficult.
accomplish.
 Artificial Intelligence Unit-6

Rule-based Systems

Rule-based systems are used as a way to store and manipulate

knowledge to interpret information in a useful way. They are often
used in artificial intelligence applications and research.

An RBS consists of a knowledge base and an inference engine. The
 knowledge base contains rules and facts.
 
 A typical rule-based system has four basic components:

A list of rules or rule base, which is a specific type of knowledge
base.
An inference engine or semantic reasoner, which infers
information or takes action based on the interaction of input and
the rule base. The interpreter executes a production system
program by performing the following match-resolve-act cycle:

o Match: In this first phase, the left-hand sides of all productions

are matched against the contents of working memory. As a
result a conflict set is obtained, which consists of instantiations
of all satisfied productions. An instantiation of a production is
an ordered list of working memory elements that satisfies the
left-hand side of the production.
o Conflict-Resolution: In this second phase, one of the
production instantiations in the conflict set is chosen for
execution. If no productions are satisfied, the interpreter halts.
o Act: In this third phase, the actions of the production selected
in the conflict-resolution phase are executed. These actions may
change the contents of working memory. At the end of this
phase, execution returns to the first phase.
Temporary working memory- set of facts.
 Artificial Intelligence Unit-6

A user interface or other connection to the outside world through

which input and output signals are received and sent.

 
 The most common RBS modes of operation are:

 forward chaining (stimulus driven)


 backward chaining (goal directed)


Forward Chaining: Forward chaining mode of operation means that a
rule is triggered when changes in the working  memory produce a
situation that matches all of its antecedents.

Forward chaining is the process of inferring then-patterns from if-
patterns that is consequents from antecedents. When an antecedent
matches an assertion the antecedent is satisfied. When all antecedents
of a rule are satisfied the rule is triggered. In deduction systems all
triggered rules are allowed and may fire.
 Artificial Intelligence Unit-6

Forward Chaining Algorithm

Repeat
For each rule do
- Match all its antecedents to
the facts from the Working memory
- if all antecedents of a rule are matched,
Execute is consequents
until no rule produces a new assertion, or the goal is satisfied.


Backward Chaining: The backward chaining mode of operation
means that the systems begins with a goal and successively
examines any rules with matching consequents. These candidate
rules are considered one at a time. The unmet conditions are in
turn reintroduced as new goals. The control procedure then shifts
attention recursively toward the new goal. The effort terminates

when the top goal is finally satisfied.
 
 In a Rule-based System:
 The domain knowledge is represented by a set of IF-THEN
production rules

 Data is represented by a set of facts about the current situation.

 The inference engine compares each rule stored in the
knowledge base with facts contained in the database.

 When the IF (condition) part of the rule matches a fact, the rule
is fired and its THEN (action) part is executed.
 Artificial Intelligence Unit-6

Rule 1: IF Y is true AND D is true THEN Z is true

Rule 2: IF X is true AND B is true AND E is true THEN Y is true
Rule 3: IF A is true THEN X is true

Blackboard System
A blackboard system is an artificial intelligence approach based on the
blackboard architectural model, where a common knowledge base, the
"blackboard", is iteratively updated by a diverse group of specialist
knowledge sources, starting with a problem specification and ending
with a solution.

Each knowledge source updates the blackboard with a partial solution
 when its internal constraints match the blackboard state. In this way,
the specialists work together to solve the problem.

The blackboard model was designed to handle complex,  ill-defined
problems, where the solution is the sum of its parts.
  A blackboard-system application consists of three major components:
 The software specialist modules, which are called knowledge
sources (KSs). Like the human experts at a blackboard, each
knowledge source provides specific expertise needed by the
 application.
 The blackboard, a shared repository of problems, partial solutions,
suggestions, and contributed information. The blackboard can be
thought of as a dynamic "library" of contributions to the current
problem that have been recently "published" by other knowledge
sources
 Artificial Intelligence Unit-6

 The control shell, which controls the flow of problem-solving

activity in the system. KSs need a mechanism to organize their use
in the most effective and coherent fashion. In a blackboard system,
this is provided by the control shell.


The advantages of a blackboard include separation of knowledge into
independent modules with each module being free to use the
appropriate  technology to arrive at the best solution with the most
efficiency.
 Artificial Intelligence Unit-6

Model-based Expert system :

In artificial intelligence, model-based reasoning refers to an inference

method used in expert systems based on a model of the physical world. With
this approach, the main focus of application development is developing the
model. Then at run time, an "engine" combines this model knowledge with
observed data to derive conclusions such as a diagnosis or a prediction.

In a model-based reasoning system knowledge can be represented using

causal rules. For example, in a medical diagnosis system the knowledge
base may contain the following rule:

There are many other forms of models that may be used. Models might be quantitative (for
instance, based on mathematical equations) or qualitative (for instance, based on cause/effect
models.) They may include representation of uncertainty. They might represent behaviour over
time. They might represent "normal" behaviour, or might only represent abnormal behaviour, as
in the case of the examples above. Model types and usage for model-based reasoning are
discussed in

Case based expert system:

Case-based reasoning (CBR), broadly construed, is the

process of solving new problems based on the solutions of similar past
problems. An auto mechanic who fixes an engine by recalling another car
that exhibited similar symptoms is using case-based reasoning. A lawyer
who advocates a particular outcome in a trial based on legal precedents or a
judge who creates case law is using case-based reasoning. So, too, an
engineer copying working elements of nature (practicing bio mimicry), is
treating nature as a database of solutions to problems. Case-based
reasoning is a prominent type of analogy solution making.

It has been argued that case-based reasoning is not only a

powerful method for computer reasoning, but also a pervasive behaviour in
everyday human problem solving; or, more radically, that all reasoning is
based on past cases personally experienced. This view is related to prototype
theory, which is most deeply explored in cognitive science.

Case-based reasoning has been formalized for purposes of computer

reasoning as a four-step process:
 Artificial Intelligence Unit-6

Retrieve: Given a target problem, retrieve from memory cases

relevant to solving it. A case consists of a problem, its solution, and,
typically, annotations about how the solution was derived. For example,
suppose Fred wants to prepare blueberry pancakes. Being a novice cook, the
most relevant experience he can recall is one in which he successfully made
plain pancakes. The procedure he followed for making the plain pancakes,
together with justifications for decisions made along the way, constitutes
Fred's retrieved case.

Reuse: Map the solution from the previous case to the target problem. This
may involve adapting the solution as needed to fit the new situation. In the
pancake example, Fred must adapt his retrieved solution to include the
addition of blueberries.

Revise: Having mapped the previous solution to the target situation, test the
new solution in the real world (or a simulation) and, if necessary, revise.
Suppose Fred adapted his pancake solution by adding blueberries to the
batter. After mixing, he discovers that the batter has turned blue – an
undesired effect. This suggests the following revision: delay the addition of
blueberries until after the batter has been ladled into the pan.

Retain: After the solution has been successfully adapted to the target
problem, store the resulting experience as a new case in memory. Fred,
accordingly, records his new-found procedure for making blueberry
pancakes, thereby enriching his set of stored experiences, and better
preparing him for future pancake-making demands.

Hybrid expert systems:

Hybrid expert systems employ two or more representations of expertise to

emulate reasoning in some domain. Typically, the component expert
systems are highly complementary. That is, the weaknesses of one are
compensated for by the other and vice versa.

Rule-based expert systems (RBES) -- which you create with Jess -- are
very good at making decisions where the domain knowledge is very well
structured and defined. Because of this, RBES can be used to reason from
first principles to arrive at a general conclusion. RBES tend not to perform
well when the domain knowledge is ill-conditioned, poorly-structured, or
sparse.

In these instances, RBES can be paired with another expert systems

metaphor to overcome that limitation. One particular metaphor that works
very well with RBES is Case-Based Reasoning (CBR). A CBR system
 Artificial Intelligence Unit-6

emulates the very human-like practice of re-using past experiences and

problem solutions to synthesize new solutions to new problems. Collections
of past problems and their associated solutions are stored as cases. When
given some search criteria, a CBR system tries to find the case or cases that
most closely match the criteria according to some evaluation algorithm. The
most popular algorithm is called Nearest Neighbor. The strength of CBR is
that if the adapted solution works, it is rolled back into the case base as new
knowledge -- therefore the more a CBR system is used, the more it learns.

Applications:

1.Diagnosis and Troubleshooting of Devices and Systems of All Kinds

MYCIN
MYCIN was an backward chaining expert system that used artificial
intelligence to identify bacteria causing severe infections, such as
bacteremia and meningitis, and to recommend antibiotics, with the
dosage adjusted for patient's body weight .

MYCIN was developed over five or six years in the early 1970s
at Stanford University. It was written in Lisp.
MYCIN operated using fairly simple inference engine, and a knowledge
base of ~600 rules. It would query the physician running the program
via a long series of simple yes/no or textual questions.
 
At the end, it provided a list of possible culprit bacteria ranked from
high to low based on the probability of each diagnosis,
its confidence in each diagnosis' probability, the reasoning behind
each diagnosis.

 Mycin is a program that diagnoses infectious diseases. It reasons 
backward from its goal of determining the cause of a patient illness.

 It attempts to solve its goal of recommending a therapy for  a particular
patient by first finding the cause of the patient’s illness.


It uses its production rules to reason backward from goals to clinical
observations.
 Artificial Intelligence Unit-6

DENDRAL
Dendral was a project in artificial intelligence (AI) of the 1960s, and
the computer software expert system that it produced. Its primary
aim was to study hypothesis formation and discovery in science.
For that, a specific task in science was chosen: help organic
chemists in identifying unknown organic molecules, by analyzing
their mass spectra and using knowledge of chemistry.

The software program Dendral is considered the first expert system
 
because it automated the decision-making process and problem-
solving behaviour of organic chemists.
It was written in the LISP programming language, which was
considered the language of AI because of its flexibility.

Dendral is a program that uses mass spectra or other experimental
data together with knowledge base of chemistry, to produce a set of
 
possible chemical structures that may be responsible for producing
the data.

A mass spectrum of a compound is produced by a mass
 
spectrometer, and is used to determine its molecular weight, the
sum of the masses of its atomic constituents.

For example, the compound water (H2O), has a molecular weight of
18 since hydrogen has a mass of 1.01 and oxygen 16.00, and its
 mass spectrum has a peak at 18 units.

Dendral uses input mass and the knowledge of atomic mass
numbers and valence rules, to determine the possible
 
combinations of atomic constituents whose mass would add up to
 18.


 Artificial Intelligence Unit-6


 complex, the
As the weight increases and the molecules become more
number of possible compounds increases drastically.

2.Planning and Scheduling

Systems that fall into this class analyze a set of one or more potentially complex
and interacting goals in order to determine a set of actions to achieve those goals,
and/or provide a detailed temporal ordering of those actions, taking into account
personnel, materiel, and other constraints. This class has great commercial
potential, which has been recognized. Examples involve airline scheduling of
flights, personnel, and gates; manufacturing job-shop scheduling; and
manufacturing process planning.

3.Configuration of Manufactured Objects from Subassemblies

Configuration, whereby a solution to a problem is synthesized from a given set of

elements related by a set of constraints, is historically one of the most important of
expert system applications. Configuration applications were pioneered by computer
companies as a means of facilitating the manufacture of semi-custom
minicomputers (McDermott 1981). The technique has found its way into use in
many different industries, for example, modular home building, manufacturing,
and other problems involving complex engineering design and manufacturing.

4.Financial Decision Making

The financial services industry has been a vigorous user of expert system
techniques. Advisory programs have been created to assist bankers in determining
whether to make loans to businesses and individuals. Insurance companies have
used expert systems to assess the risk presented by the customer and to determine a
price for the insurance. A typical application in the financial markets is in foreign
exchange trading.

5.Knowledge Publishing

This is a relatively new, but also potentially explosive area. The primary function
of the expert system is to deliver knowledge that is relevant to the user's problem,
in the context of the user's problem. The two most widely distributed expert
systems in the world are in this category. The first is an advisor which counsels a
user on appropriate grammatical usage in a text. The second is a tax advisor that
accompanies a tax preparation program and advises the user on tax strategy,
tactics, and individual tax policy.
 Artificial Intelligence Unit-6

6.Process Monitoring and Control

Systems falling in this class analyze real-time data from physical devices with the
goal of noticing anomalies, predicting trends, and controlling for both optimality
and failure correction. Examples of real-time systems that actively monitor
processes can be found in the steel making and oil refining industries.

7.Design and Manufacturing

These systems assist in the design of physical devices and processes, ranging from
high-level conceptual design of abstract entities all the way to factory floor
configuration of manufacturing processes.

8.R1

The R1 (internally called XCON, for eXpert CONfigurer) program was a

production-rule-based system written in OPS5 in 1978 to assist in the
ordering of DEC's VAX computer systems by automatically selecting
the computer system components based on the customer's
requirements.

R1 is a program that configures VAX-11/780 computer systems.
Given a customer's order, it determines what, if any, modifications
have to be made to the order for reasons of system functionality and
 
produces a number of diagrams showing how the various components
on the order are to be associated.

The program is currently being used on a regular basis by Digital
 
Equipment Corporation's manufacturing organization. R1 is
implemented as a production system.

It uses Match as its principal problem solving method; it has sufficient
knowledge of the configuration domain and of the peculiarities of the
various configuration constraints that at each step in the
configuration process, it simply recognizes what to do. Consequently,
 little search is required in order for it to configure a computer system.

A typical VAX system includes many components from 50 to 150 in
the following categories: CPU, memory control units, peripherals: tape

drives, floppy disks, hard disks,  size, printers, drivers for the
peripherals, cabinets, and cables.

There are many constraints because most permutations of
components are not feasible. Only certain components can be

attached to one another, and this limits the possible combinations
that can be considered.

Knowledge about the interconnections is represented by thousands of
production rules. The control strategy is hierarchical and sub-
 Artificial Intelligence Unit-6

problems are solved in order of the importance of their associated

goal.
 
XCON uses forward reasoning and is written in the OPS5 language.

9.PROSPECTOR


PROSPECTOR is an expert system designed for decision-making
problems in mineral exploration. It aids geologists in evaluating the
  site or region for occurrences of ore
favourability of an exploration
deposits of particular types.

 
Once a site has been identified, PROSPECTOR can also be used for
drilling-site selection.
 PROSPECTOR is written in INTERLISP, an advanced dialect of the
 LISP language.

PROSPECTOR can reach a conclusion about a particular ore deposit.
 It gives a certainty value of the ore deposit. It as well provides the
explanation text for the conclusion.

When PROSPECTOR is used for the drilling  site selection, it could also
produce the favourability map of the site.

Assignment-Cum-Tutorial Questions

I) Objective Questions
1. Define the term “Expert System”.
2. The components of expert system are _________________,
__________________and _________________.
3. Knowledge comprises of _____________.

(a) Data (b) Information (c) Past Experience (d)

All the above
4. The information that is widely accepted by the Knowledge Engineers
and scholars in the task domain is called __________ knowledge.
(a) Factual (b) Heuristic (c) Domain (d) none
5. Knowledge that is about practice, accurate judgment, one’s ability of
evaluation, and guessing is called _______________ knowledge.
(a) Factual (b) Heuristic (c) Domain (d) none
 Artificial Intelligence Unit-6

6. ____________categorizes and organizes the information in a meaningful

way.
(a) Knowledge Engineer (b) Human Expert (c) User (d) Tool
7. ___________is a strategy of an expert system to answer the question,
“What can happen next?”
(a) Forward Chaining (b) Backward chaining (c) both
(d) none
8. ___________is a strategy of an expert system finds out the answer to
the question, “Why this happened?”
(a) Forward Chaining (b) Backward chaining (c) both (d) none
9. ____________ is an expert system without knowledge base.
(a) Tool (b) Shell (c) Task (d) none
10. In an Expert System, the entire problem related expertise is encoded
in _______________________.

(i) Data Structures

(ii) (ii) Programs
(a) Only (i) (b) Only (ii) (c) Both (i) and (ii) (d) none
11. In a traditional system, the entire problem related expertise is
encoded in _______________________.

(i) Data Structures

(ii) (ii) Programs
(a) Only (i) (b) Only (ii) (c) Both (i) and (ii) (d) none
12. The knowledgebase in a Rule-base system consists of _________.
(a) Rules (b) Facts (c) Both a & b (d) productions
13. MYCIN is a ______________ expert system.
(a) Forward Chaining (b) Backward chaining (c) both (d) none
14. DENDRAL was written in the __________ programming language. (a)
PROLOG (b) LISP (c) FORTRAN (d) PYTHON
 Artificial Intelligence Unit-6

II) Descriptive Questions

1. List the characteristics and capabilities of Expert System.

2. Explain the components of an expert system.
3. Distinguish between Forward chaining and Backward chaining.
4. Describe the phases of developing an Expert system.
5. What do you mean by expert system technology? Explain.
6. Distinguish Expert system and Traditional system.
7. Explain about Rule-based Systems.
8. Explain about case based expert system.
9. Explain about model based expert system
10. Discuss about hybrid expert system.
11. List out the expert system applications.
12. Write short notes on:
(i) MYCIN

(ii) DENDRAl

(iii) R1