Relational Database Design

Chapter 7: Relational Database

Design

 First Normal Form

 Pitfalls in Relational Database Design
 Functional Dependencies
 Decomposition
 Boyce-Codd Normal Form
 Third Normal Form
 Multivalued Dependencies and Fourth Normal Form
 Overall Database Design Process
 First Normal Form

 Domain is atomic if its elements are considered to

be indivisible units
 Examples of non-atomic domains:
 Set of names, composite attributes
 Identification numbers like CS101 that can be
broken up into parts
 A relational schema R is in first normal form if the
domains of all attributes of R are atomic
 Non-atomic values complicate storage and
encourage redundant (repeated) storage of data
 E.g. Set of accounts stored with each customer, and
set of owners stored with each account
 We assume all relations are in first normal form (revisit
this in Chapter 9 on Object Relational Databases)
 First Normal Form (Contd.)
 Atomicity is actually a property of how the elements
of the domain are used.
 E.g. Strings would normally be considered indivisible
 Suppose that students are given roll numbers which
are strings of the form CS0012 or EE1127
 If the first two characters are extracted to find the
department, the domain of roll numbers is not atomic.
 Doing so is a bad idea: leads to encoding of
information in application program rather than in the
database.
 Pitfalls in Relational Database
Design

 Relational database design requires that we

find a “good” collection of relation schemas.
A bad design may lead to
 Repetition of Information.
 Inability to represent certain information.
 Design Goals:
 Avoid redundant data
 Ensure that relationships among attributes are
represented
 Facilitate the checking of updates for violation
of database integrity constraints.
 Example
 Consider the relation schema:
Lending-schema = (branch-name, branch-city, assets,
customer-name, loan-number,
amount)

 Redundancy:
 Data for branch-name, branch-city, assets are repeated for each
loan that a branch makes
 Wastes space
 Complicates updating, introducing possibility of inconsistency of
assets value
 Null values
 Cannot store information about a branch if no loans exist
 Can use null values, but they are difficult to handle.
 Decomposition

 Decompose the relation schema Lending-schema

into:
Branch-schema = (branch-name, branch-city,assets)
Loan-info-schema = (customer-name, loan-number,
branch-name,
amount)
 All attributes of an original schema (R) must
appear in the decomposition (R1, R2):
R = R1  R2
 Lossless-join decomposition.
For all possible relations r on schema R
r = R1 (r) R2 (r)
 Example of Non Lossless-Join
Decomposition

 Decomposition of R = (A, B)
R2 = (A) R2 = (B)

A B A B

 1  1
 2  2
 1 B(r)
A(r)
r
A B
A (r) B (r)
 1
 2
 1
 2
 Goal — Devise a Theory for the
Following

 Decide whether a particular relation R is in “good”

form.
 In the case that a relation R is not in “good” form,
decompose it into a set of relations {R1, R2, ..., Rn}
such that
 each relation is in good form
 the decomposition is a lossless-join decomposition
 Our theory is based on:
 functional dependencies
 multivalued dependencies
 Functional Dependencies

 Constraints on the set of legal relations.

 Require that the value for a certain set of attributes
determines uniquely the value for another set of
attributes.
 A functional dependency is a generalization of the
notion of a key.
 Functional Dependencies (Cont.)
 Let R be a relation schema
  R and   R
 The functional dependency

holds on R if and only if for any legal relations r(R),
whenever any two tuples t1 and t2 of r agree on the
attributes , they also agree on the attributes . That is,
t1[] = t2 []  t1[ ] = t2 [ ]
 Example: Consider r(A,B) with the following instance of
r.
1 4
1 5
3 7

 On this instance, A  B does NOT hold, but B  A does

hold.
 Functional Dependencies (Cont.)

 K is a superkey for relation schema R if and only if K 

R
 K is a candidate key for R if and only if
 K  R, and
 for no   K,   R
 Functional dependencies allow us to express
constraints that cannot be expressed using superkeys.
Consider the schema:
Loan-info-schema = (customer-name, loan-
number,
branch-name, amount).
We expect this set of functional dependencies to hold:
loan-number  amount
loan-number  branch-name
but would not expect the following to hold:
loan-number  customer-name
 Use of Functional Dependencies

 We use functional dependencies to:

 test relations to see if they are legal under a given set of
functional dependencies.
r is legal under a set F of functional
 If a relation
dependencies, we say that r satisfies F.
 specify constraints on the set of legal relations
 We say that F holds on R if all legal relations on R
satisfy the set of functional dependencies F.
 Note: A specific instance of a relation schema may
satisfy a functional dependency even if the functional
dependency does not hold on all legal instances. For
example, a specific instance of Loan-schema may, by
chance, satisfy
loan-number  customer-name.
 Functional Dependencies (Cont.)

 A functional dependency is trivial if it is satisfied by

all instances of a relation
 E.g.
 customer-name, loan-number  customer-name
 customer-name  customer-name
 In general,    is trivial if   
 Closure of a Set of Functional
Dependencies
 Given a set F set of functional dependencies, there are
certain other functional dependencies that are logically
implied by F.
 E.g. If A  B and B  C, then we can infer that A  C
 The set of all functional dependencies logically implied by
F is the closure of F.
 We denote the closure of F by F+.
 We can find all of F+ by applying Armstrong’s Axioms:
 if   , then    (reflexivity)
 if   , then      (augmentation)
 if   , and   , then    (transitivity)
 These rules are
 sound (generate only functional dependencies that actually
hold) and
 complete (generate all functional dependencies that hold).
 Example
 R = (A, B, C, G, H, I)
F={ AB
AC
CG  H
CG  I
B  H}
 some members of F+
A  H
 by transitivity from A  B and B  H
 AG  I
 by augmenting A  C with G, to get AG  CG
and then transitivity with CG  I
 CG  HI
 from CG  H and CG  I : “union rule” can be inferred
from
– definition of functional dependencies, or
– Augmentation of CG  I to infer CG  CGI,
augmentation of
CG  H to infer CGI  HI, and then transitivity
 Procedure for Computing F+

 To compute the closure of a set of functional dependencies

F:

F+ = F
repeat
for each functional dependency f in F+
apply reflexivity and augmentation rules on f
add the resulting functional dependencies to F+
for each pair of functional dependencies f1and f2 in F+
if f1 and f2 can be combined using transitivity
then add the resulting functional dependency
to F+
until F+ does not change any further

NOTE: We will see an alternative procedure for this task later

 Closure of Functional
Dependencies (Cont.)

 We can further simplify manual computation of

F+ by using the following additional rules.
 If    holds and    holds, then     holds
(union)
 If     holds, then    holds and    holds
(decomposition)
 If    holds and     holds, then     holds
(pseudotransitivity)
The above rules can be inferred from Armstrong’s
axioms.
 Closure of Attribute Sets

 Given a set of attributes  define the closure of 

under F (denoted by +) as the set of attributes that
are functionally determined by  under F:
   is in F+    +
 Algorithm to compute +, the closure of  under F

result := ;
while (changes to result) do
for each    in F do
begin
if   result then result := result  
end
Example of Attribute Set Closure
 R = (A, B, C, G, H, I)
 F = {A  B
AC
CG  H
CG  I
B  H}
 (AG)+
1. result = AG
2. result = ABCG (A  C and A  B)
3. result = ABCGH (CG  H and CG  AGBC)
4. result = ABCGHI (CG  I and CG  AGBCH)
 Is AG a candidate key?
1. Is AG a super key?
1. Does AG  R?
2. Is any subset of AG a superkey?
1. Does A+  R?
2. Does G+  R?
 Uses of Attribute Closure
There are several uses of the attribute closure
algorithm:
 Testing for superkey:
 To test if  is a superkey, we compute +, and check if
+ contains all attributes of R.
 Testing functional dependencies
 To check if a functional dependency    holds (or, in
other words, is in F+), just check if   +.
 That is, we compute + by using attribute closure, and
then check if it contains .
 Is a simple and cheap test, and very useful
 Computing closure of F
 For each   R, we find the closure +, and for each S 
+, we output a functional dependency   S.
 Canonical Cover

 Sets of functional dependencies may have

redundant dependencies that can be inferred from
the others
 Eg: A  C is redundant in: {A  B, B  C, A  C}
 Parts of a functional dependency may be redundant
 E.g. on RHS: {A  B, B  C, A  CD} can be
simplified to
{A  B, B  C, A  D}
 E.g. on LHS: {A  B, B  C, AC  D} can be
simplified to
{A  B, B  C, A  D}
 Intuitively, a canonical cover of F is a “minimal” set
of functional dependencies equivalent to F, with no
redundant dependencies or having redundant parts
of dependencies
 Extraneous Attributes
 Consider a set F of functional dependencies and
the functional dependency    in F.
 Attribute A is extraneous in  if A  
and F logically implies (F – {  })  {( – A)  }.
 Attribute A is extraneous in  if A  
and the set of functional dependencies
(F – {  })  { ( – A)} logically implies F.
 Note: implication in the opposite direction is trivial
in each of the cases above, since a “stronger”
functional dependency always implies a weaker
one
 Example: Given F = {A  C, AB  C }
 B is extraneous in AB  C because A  C logically
implies
AB  C.
 Example: Given F = {A  C, AB  CD}
 C is extraneous in AB  CD since A  C can be
inferred even after deleting C
 Testing if an Attribute is
Extraneous

 Consider a set F of functional dependencies and the

functional dependency    in F.
 To test if attribute A   is extraneous in 
1. compute (A – {})+ using the dependencies in F
2. check that (A – {})+ contains ; if it does, A is
extraneous
 To test if attribute A   is extraneous in 
1. compute + using only the dependencies in
F’ = (F – {  })  { ( – A)},
2. check that + contains A; if it does, A is
extraneous
 Canonical Cover

 A canonical cover for F is a set of dependencies F such that

c
 F logically implies all dependencies in Fc, and
 Fc logically implies all dependencies in F, and
 No functional dependency in Fc contains an extraneous attribute,
and
 Each left side of functional dependency in Fc is unique.
 To compute a canonical cover for F:
repeat
Use the union rule to replace any dependencies in F
1  1 and 1  1 with 1  1 2
Find a functional dependency    with an
extraneous attribute either in  or in 
If an extraneous attribute is found, delete it from   
until F does not change
 Note: Union rule may become applicable after some
extraneous attributes have been deleted, so it has to be re-
applied
Example of Computing a Canonical
Cover

 R = (A, B, C)
F = {A  BC
BC
AB
AB  C}
 Combine A  BC and A  B into A  BC
 Set is now {A  BC, B  C, AB  C}
 A is extraneous in AB  C because B  C logically
implies AB  C.
 Set is now {A  BC, B  C}
 C is extraneous in A  BC since A  BC is logically
implied by A  B and B  C.
 The canonical cover is:
AB
BC
 Goals of Normalization
 Decide whether a particular relation R is in “good” form.
 In the case that a relation R is not in “good” form,
decompose it into a set of relations {R1, R2, ..., Rn} such
that
 each relation is in good form
 the decomposition is a lossless-join decomposition
 Our theory is based on:
 functional dependencies
 multivalued dependencies
 Decomposition

 Decompose the relation schema Lending-schema into:

Branch-schema = (branch-name, branch-city,assets)
Loan-info-schema = (customer-name, loan-number,
branch-name, amount)
 All attributes of an original schema (R) must appear in
the decomposition (R1, R2):
R = R1  R2
 Lossless-join decomposition.
For all possible relations r on schema R
r = R1 (r) R2 (r)
 A decomposition of R into R1 and R2 is lossless join if
and only if at least one of the following dependencies is
in F+:
 R1  R2  R1
 R1  R2  R2
Example of Lossy-Join Decomposition

 Lossy-join decompositions result in information loss.

 Example: Decomposition of R = (A, B)
R2 = (A) R2 = (B)

A B A B

 1  1
 2  2
 1 B(r)
A(r)
r
A B
A (r) B (r)
 1
 2
 1
 2
Normalization Using Functional Dependencie

 When we decompose a relation schema R

with a set of functional dependencies F into
R1, R2,.., Rn we want
 Lossless-join decomposition: Otherwise decomposition
would result in information loss.
 No redundancy: The relations Ri preferably should be in
either Boyce-Codd Normal Form or Third Normal Form.
 Dependency preservation: Let Fi be the set of
dependencies F+ that include only attributes in Ri.
 Preferably the decomposition should be dependency
preserving, that is, (F1  F2  …  Fn)+ = F+
 Otherwise, checking updates for violation of functional
dependencies may require computing joins, which is
expensive.
 Example

 R = (A, B, C)
F = {A  B, B  C)
 R1 = (A, B), R2 = (B, C)
 Lossless-join decomposition:
R1  R2 = {B} and B  BC
 Dependency preserving
 R1 = (A, B), R2 = (A, C)
 Lossless-join decomposition:
R1  R2 = {A} and A  AB
 Not dependency preserving
(cannot check B  C without computing R1 R2 )
 Testing for Dependency
Preservation

 To check if a dependency  is preserved in a

decomposition of R into R1, R2, …, Rn we apply the
following simplified test (with attribute closure done w.r.t.
F)
 result = 
while (changes to result) do
for each Ri in the decomposition
t = (result  Ri)+  Ri
result = result  t
 If result contains all attributes in , then the functional
dependency
   is preserved.
 We apply the test on all dependencies in F to check if a
decomposition is dependency preserving
 This procedure takes polynomial time, instead of the
exponential time required to compute F+ and (F1  F2  …
 Fn ) +
 Boyce-Codd Normal Form

A relation schema R is in BCNF with respect to a set F of functiona

dependencies if for all functional dependencies in F+ of the form
 , where   R and   R, at least one of the following holds:

    is trivial (i.e.,   )
  is a superkey for R
 Example

 R = (A, B, C)
F = {A  B
B  C}
Key = {A}
 R is not in BCNF
 Decomposition R1 = (A, B), R2 = (B, C)
 R1 and R2 in BCNF
 Lossless-join decomposition
 Dependency preserving
 Testing for BCNF

 To check if a non-trivial dependency  causes a violation of

BCNF
1. compute + (the attribute closure of ), and
2. verify that it includes all attributes of R, that is, it is a superkey of
R.
 Simplified test: To check if a relation schema R with a given
set of functional dependencies F is in BCNF, it suffices to
check only the dependencies in the given set F for violation of
BCNF, rather than checking all dependencies in F+.
 We can show that if none of the dependencies in F causes a
violation of BCNF, then none of the dependencies in F+ will cause a
violation of BCNF either.
 However, using only F is incorrect when testing a relation in a
decomposition of R
 E.g. Consider R (A, B, C, D), with F = { A B, B C}
 Decompose R into R1(A,B) and R2(A,C,D)
 Neither of the dependencies in F contain only attributes from
(A,C,D) so we might be mislead into thinking R2 satisfies BCNF.
 In fact, dependency A  C in F+ shows R2 is not in BCNF.
 BCNF Decomposition Algorithm
result := {R};
done := false;
compute F+;
while (not done) do
if (there is a schema Ri in result that is not in
BCNF)
then begin
let    be a nontrivial functional
dependency that holds on Ri
such that   Ri is not in F+,
and    = ;
result := (result – Ri)  (Ri – )  (,  );
end
else done := true;
Note: each Ri is in BCNF, and decomposition is lossless-
join.
Example of BCNF Decomposition
 R = (branch-name, branch-city, assets,
customer-name, loan-number, amount)
F = {branch-name  assets branch-city
loan-number  amount branch-name}
Key = {loan-number, customer-name}
 Decomposition
 R1 = (branch-name, branch-city, assets)
 R2 = (branch-name, customer-name, loan-number,
amount)
 R3 = (branch-name, loan-number, amount)
 R4 = (customer-name, loan-number)
 Final decomposition
R 1, R 3, R 4
 Testing Decomposition for BCNF

 To check if a relation Ri in a decomposition of R is in BCNF,

 Either test Ri for BCNF with respect to the restriction of F to Ri
(that is, all FDs in F+ that contain only attributes from Ri)
 or use the original set of dependencies F that hold on R, but
with the following test:
– for every set of attributes   Ri, check that + (the
attribute closure of ) either includes no attribute of Ri- ,
or includes all attributes of Ri.
 If the condition is violated by some   in F, the
dependency
 (+ - )  Ri
can be shown to hold on Ri, and Ri violates BCNF.
 We use above dependency to decompose Ri
 BCNF and Dependency
Preservation

It is not always possible to get a BCNF decomposition that is

dependency preserving
 R = (J, K, L)
F = {JK  L
L  K}
Two candidate keys = JK and JL
 R is not in BCNF
 Any decomposition of R will fail to preserve

JK  L
 Third Normal Form: Motivation

 There are some situations where

 BCNF is not dependency preserving, and
 efficient checking for FD violation on updates is
important
 Solution: define a weaker normal form, called Third
Normal Form.
 Allows some redundancy (with resultant problems; we
will see examples later)
 But FDs can be checked on individual relations without
computing a join.
 There is always a lossless-join, dependency-preserving
decomposition into 3NF.
 Third Normal Form

 A relation schema R is in third normal form (3NF) if

for all:
   in F+
at least one of the following holds:
    is trivial (i.e.,   )
  is a superkey for R
 Each attribute A in  –  is contained in a candidate key
for R.
(NOTE: each attribute may be in a different candidate
key)
 If a relation is in BCNF it is in 3NF (since in BCNF
one of the first two conditions above must hold).
 Third condition is a minimal relaxation of BCNF to
ensure dependency preservation (will see why later).
 3NF (Cont.)

 Example
 R = (J, K, L)
F = {JK  L, L  K}
 Two candidate keys: JK and JL
 R is in 3NF
JK  L JK is a superkey
LK K is contained in a candidate key
 BCNF decomposition has (JL) and (LK)
 Testing for JK  L requires a join
 There is some redundancy in this schema
 Equivalent to example in book:
Banker-schema = (branch-name, customer-name, banker-
name)
banker-name  branch name
branch name customer-name  banker-name
 Testing for 3NF

 Optimization: Need to check only FDs in F, need not

check all FDs in F+.
 Use attribute closure to check, for each dependency
  , if  is a superkey.
 If  is not a superkey, we have to verify if each
attribute in  is contained in a candidate key of R
 this test is rather more expensive, since it involve
finding candidate keys
 testing for 3NF has been shown to be NP-hard
 Interestingly, decomposition into third normal form
(described shortly) can be done in polynomial time
 3NF Decomposition Algorithm
Let Fc be a canonical cover for F;
i := 0;
for each functional dependency    in Fc do
if none of the schemas Rj, 1  j  i contains  
then begin
i := i + 1;
Ri :=  
end
if none of the schemas Rj, 1  j  i contains a candidate
key for R
then begin
i := i + 1;
Ri := any candidate key for R;
end
return (R1, R2, ..., Ri)
 3NF Decomposition Algorithm
(Cont.)

 Above algorithm ensures:

 each relation schema Ri is in 3NF
 decomposition is dependency preserving and lossless-join
 Proof of correctness is at end of this file (click here)
 Example

 Relation schema:
Banker-info-schema = (branch-name, customer-name,
banker-name, office-number)
 The functional dependencies for this relation
schema are:
banker-name  branch-name office-number
customer-name branch-name  banker-name
 The key is:

{customer-name, branch-name}
 Applying 3NF to Banker-info-
schema

 The for loop in the algorithm causes us to include

the following schemas in our decomposition:
Banker-office-schema = (banker-name,
branch-name,
office-number)
Banker-schema = (customer-name, branch-
name,
banker-name)

 Since Banker-schema contains a candidate key for

Banker-info-schema, we are done with the
decomposition process.
 Comparison of BCNF and 3NF

 It is always possible to decompose a relation into

relations in 3NF and
 the decomposition is lossless
 the dependencies are preserved
 It is always possible to decompose a relation into
relations in BCNF and
 the decomposition is lossless
 it may not be possible to preserve dependencies.
 Comparison of BCNF and 3NF
(Cont.)

 Example of problems due to redundancy in 3NF

 R = (J, K, L)
F = {JK  L, L  K}
J L K
j1 l1 k1

j2 l1 k1

j3 l1 k1
null l2 k2
A schema that is in 3NF but not in BCNF has the problems of
 repetition of information (e.g., the relationship l1, k1)
 need to use null values (e.g., to represent the relationship
l2, k2 where there is no corresponding value for J).
 Design Goals

 Goal for a relational database design is:

 BCNF.
 Lossless join.
 Dependency preservation.
 If we cannot achieve this, we accept one of
 Lack of dependency preservation
 Redundancy due to use of 3NF
 Interestingly, SQL does not provide a direct way of
specifying functional dependencies other than
superkeys.
Can specify FDs using assertions, but they are
expensive to test
 Even if we had a dependency preserving decomposition,
using SQL we would not be able to efficiently test a
functional dependency whose left hand side is not a key.
 Testing for FDs Across Relations

 If decomposition is not dependency preserving, we can

have an extra materialized view for each dependency 
 in Fc that is not preserved in the decomposition
 The materialized view is defined as a projection on  
of the join of the relations in the decomposition
 Many newer database systems support materialized
views and database system maintains the view when
the relations are updated.
 No extra coding effort for programmer.
 The FD becomes a candidate key on the materialized
view.
 Space overhead: for storing the materialized view
 Time overhead: Need to keep materialized view up to
date when relations are updated
 Multivalued Dependencies
 There are database schemas in BCNF that do not
seem to be sufficiently normalized
 Consider a database

classes(course, teacher, book)

such that (c,t,b)  classes means that t is qualified
to teach c, and b is a required textbook for c
 The database is supposed to list for each course the
set of teachers any one of which can be the
course’s instructor, and the set of books, all of
which are required for the course (no matter who
teaches it).
 course teacher book
database Avi DB Concepts
database Avi Ullman
database Hank DB Concepts
database Hank Ullman
database Sudarshan DB Concepts
database Sudarshan Ullman
operating systems
Avi OS Concepts
operating systems
Avi Shaw
operating systems
Jim OS Concepts
operating systems
Jim Shaw
classes
 Since there are non-trivial dependencies, (course,
teacher, book) is the only key, and therefore the
relation is in BCNF
 Insertion anomalies – i.e., if Sara is a new teacher that
can teach database, two tuples need to be inserted
(database, Sara, DB Concepts)
(database, Sara, Ullman)
 Therefore, it is better to decompose classes into:

course teacher
database Avi
database Hank
database Sudarshan
operating systems Avi
operating systems Jim
teaches
course book
database DB Concepts
database Ullman
operating systems OS Concepts
operating systems Shaw
text
We shall see that these two relations are in
Fourth Normal Form (4NF)
Multivalued Dependencies (MVDs)

 Let R be a relation schema and let   R and

  R. The multivalued dependency
  
holds on R if in any legal relation r(R), for all
pairs for tuples t1 and t2 in r such that t1[] =
t2 [], there exist tuples t3 and t4 in r such
that:
t1[] = t2 [] = t3 [] t4 []
t3[] = t1 []
t3[R – ] = t2[R – ]
t4 ] = t2[]
t4[R – ] = t1[R – ]
 MVD (Cont.)

 Tabular representation of   
 Example

 Let R be a relation schema with a set of attributes

that are partitioned into 3 nonempty subsets.
Y, Z, W
 We say that Y  Z (Y multidetermines Z)
if and only if for all possible relations r(R)
< y1, z1, w1 >  r and < y2, z2, w2 >  r
then
< y1, z1, w2 >  r and < y1, z2, w1 >  r
 Note that since the behavior of Z and W are
identical it follows that Y  Z if Y  W
 Example (Cont.)

 In our example:
course  teacher
course  book
 The above formal definition is supposed to
formalize the notion that given a particular
value of Y (course) it has associated with it a
set of values of Z (teacher) and a set of
values of W (book), and these two sets are in
some sense independent of each other.
 Note:
 If Y  Z then Y  Z
 Indeed we have (in above notation) Z1 = Z2
The claim follows.
Use of Multivalued Dependencies

 We use multivalued dependencies in two ways:

1. To test relations to determine whether they are
legal under a given set of functional and
multivalued dependencies
2. To specify constraints on the set of legal
relations. We shall thus concern ourselves only
with relations that satisfy a given set of
functional and multivalued dependencies.
 If a relation r fails to satisfy a given multivalued
dependency, we can construct a relations r
that does satisfy the multivalued dependency
by adding tuples to r.
 Theory of MVDs

 From the definition of multivalued dependency, we can

derive the following rule:
 If   , then   
That is, every functional dependency is also a
multivalued dependency
 The closure D+ of D is the set of all functional and
multivalued dependencies logically implied by D.
 We can compute D+ from D, using the formal definitions of
functional dependencies and multivalued dependencies.
 We can manage with such reasoning for very simple
multivalued dependencies, which seem to be most
common in practice
 For complex dependencies, it is better to reason about
sets of dependencies using a system of inference rules
(see Appendix C).
 Fourth Normal Form

 A relation schema R is in 4NF with respect to a set

D of functional and multivalued dependencies if for
all multivalued dependencies in D+ of the form  
, where   R and   R, at least one of the
following hold:
   is trivial (i.e.,    or    = R)
  is a superkey for schema R
 If a relation is in 4NF it is in BCNF
 Restriction of Multivalued
Dependencies

 The restriction of D to Ri is the set Di consisting of

 All functional dependencies in D+ that include only
attributes of Ri
 All multivalued dependencies of the form
  (  Ri)
where   Ri and    is in D+
 4NF Decomposition Algorithm

result: = {R};
done := false;
compute D+;
Let Di denote the restriction of D+ to Ri
while (not done)
if (there is a schema Ri in result that is not in 4NF)
then
begin
let    be a nontrivial multivalued
dependency that holds
on Ri such that   Ri is not in Di, and ;
result := (result - Ri)  (Ri - )  (, );
end
else done:= true;
Note: each Ri is in 4NF, and decomposition is lossless-join
 Example
 R =(A, B, C, G, H, I)
F ={ A  B
B  HI
CG  H }
 R is not in 4NF since A  B and A is not a superkey for R
 Decomposition
a) R1 = (A, B) (R1 is in 4NF)
b) R2 = (A, C, G, H, I) (R2 is not in 4NF)
c) R3 = (C, G, H) (R3 is in 4NF)
d) R4 = (A, C, G, I) (R4 is not in 4NF)
 Since A  B and B  HI, A  HI, A  I
e) R5 = (A, I) (R5 is in 4NF)
f)R6 = (A, C, G) (R6 is in 4NF)
 Further Normal Forms

 join dependencies generalize multivalued

dependencies
 lead to project-join normal form (PJNF) (also called
fifth normal form)
 A class of even more general constraints, leads to a
normal form called domain-key normal form.
 Problem with these generalized constraints: i hard
to reason with, and no set of sound and complete
set of inference rules.
 Hence rarely used
 Overall Database Design Process

 We have assumed schema R is given

 R could have been generated when converting E-R
diagram to a set of tables.
 R could have been a single relation containing all
attributes that are of interest (called universal relation).
 Normalization breaks R into smaller relations.
 R could have been the result of some ad hoc design of
relations, which we then test/convert to normal form.
 ER Model and Normalization

 When an E-R diagram is carefully designed, identifying

all entities correctly, the tables generated from the E-R
diagram should not need further normalization.
 However, in a real (imperfect) design there can be FDs
from non-key attributes of an entity to other attributes of
the entity
 E.g. employee entity with attributes department-number
and department-address, and an FD department-number
 department-address
 Good design would have made department an entity
 FDs from non-key attributes of a relationship set
possible, but rare --- most relationships are binary
 Universal Relation Approach

 Dangling tuples – Tuples that “disappear” in computing a join.

 Let r1 (R1), r2 (R2), …., rn (Rn) be a set of relations
 A tuple r of the relation ri is a dangling tuple if r is not in the
relation:
Ri (r1 r2 … rn)
 The relation r1 r2… rn is called a universal relation since
it involves all the attributes in the “universe” defined by
R1  R2  …  Rn
 If dangling tuples are allowed in the database, instead of
decomposing a universal relation, we may prefer to synthesize
a collection of normal form schemas from a given set of
attributes.
 Universal Relation Approach

 Dangling tuples may occur in practical database

applications.
 They represent incomplete information
 E.g. may want to break up information about loans
into:
(branch-name, loan-number)
(loan-number, amount)
(loan-number, customer-name)
 Universal relation would require null values, and
have dangling tuples
 Universal Relation Approach
(Contd.)

 A particular decomposition defines a restricted form of

incomplete information that is acceptable in our
database.
 Above decomposition requires at least one of customer-
name, branch-name or amount in order to enter a loan
number without using null values
 Rules out storing of customer-name, amount without an
appropriate loan-number (since it is a key, it can't be null
either!)
 Universal relation requires unique attribute names
unique role assumption
 e.g. customer-name, branch-name
 Reuse of attribute names is natural in SQL since relation
names can be prefixed to disambiguate names
 Denormalization for Performance

 May want to use non-normalized schema for performance

 E.g. displaying customer-name along with account-
number and balance requires join of account with
depositor
 Alternative 1: Use denormalized relation containing
attributes of account as well as depositor with all above
attributes
 faster lookup
 Extra space and extra execution time for updates
 extra coding work for programmer and possibility of error in
extra code
 Alternative 2: use a materialized view defined as
account depositor
 Benefits and drawbacks same as above, except no extra
coding work for programmer and avoids possible errors
 Other Design Issues

 Some aspects of database design are not caught by

normalization
 Examples of bad database design, to be avoided:
Instead of earnings(company-id, year, amount), use
 earnings-2000, earnings-2001, earnings-2002, etc., all
on the schema (company-id, earnings).
 Above are in BCNF, but make querying across years
difficult and needs new table each year
 company-year(company-id, earnings-2000, earnings-
2001, earnings-2002)
 Also in BCNF, but also makes querying across years
difficult and requires new attribute each year.
 Is an example of a crosstab, where values for one
attribute become column names
 Used in spreadsheets, and in data analysis tools
Proof of Correctness of 3NF
Decomposition Algorithm
Correctness of 3NF Decomposition
Algorithm
 3NF decomposition algorithm is dependency preserving
(since there is a relation for every FD in F c)
 Decomposition is lossless join
 A candidate key (C) is in one of the relations Ri in
decomposition
 Closure of candidate key under Fc must contain all attributes in
R.
 Follow the steps of attribute closure algorithm to show there is
only one tuple in the join result for each tuple in Ri
Correctness of 3NF Decomposition
Algorithm (Contd.)
Claim: if a relation Ri is in the decomposition
generated by the
above algorithm, then Ri satisfies 3NF.
 Let Ri be generated from the dependency  
 Let   be any non-trivial functional dependency on
Ri. (We need only consider FDs whose right-hand
side is a single attribute.)
 Now, B can be in either  or  but not in both.
Consider each case separately.
Correctness of 3NF Decomposition
(Contd.)
 Case 1: If B in :
 If  is a superkey, the 2nd condition of 3NF is satisfied
 Otherwise  must contain some attribute not in 
 Since   B is in F+ it must be derivable from Fc, by
using attribute closure on .
 Attribute closure not have used   - if it had been
used,  must be contained in the attribute closure of ,
which is not possible, since we assumed  is not a
superkey.
 Now, using  (- {B}) and   B, we can derive  B
(since    , and    since   B is non-trivial)
 Then, B is extraneous in the right-hand side of  ;
which is not possible since   is in Fc.
 Thus, if B is in  then  must be a superkey, and the
second condition of 3NF must be satisfied.
Correctness of 3NF Decomposition
(Contd.)
 Case 2: B is in .
 Since  is a candidate key, the third alternative in the
definition of 3NF is trivially satisfied.
 In fact, we cannot show that  is a superkey.
 This shows exactly why the third alternative is present
in the definition of 3NF.
Q.E.D.
End of Chapter
Sample lending Relation
Sample Relation r
The customer Relation
The loan Relation
The branch Relation
The Relation branch-customer
The Relation customer-loan
The Relation branch-customer customer-
loan
An Instance of Banker-schema
Tabular Representation of 
Relation bc: An Example of Reduncy in a BCNF
Relation
An Illegal bc Relation
Decomposition of loan-info
Relation of Exercise 7.4