VIVA Q&A:

IMPORTANT Q&A:

 What is Natural Language Processing (NLP)?

NLP is a field of AI that enables computers to understand, interpret, and generate human language.

 Mention any two applications of NLP.

Machine Translation (e.g., Google Translate) and Sentiment Analysis.

 What are the key challenges in NLP?

Ambiguity, context understanding, and language variability.

 Define Tokenization.
Tokenization is the process of breaking text into smaller units like words or sentences.

 What is the difference between Syntax and Semantics in NLP?

Syntax deals with the structure of sentences, while semantics focuses on the meaning.

 What is a Corpus in NLP?

A corpus is a large collection of text used for training NLP models.

 Explain the term Morphology in NLP.

Morphology is the study of the structure and formation of words.

 What is Lemmatization?
Lemmatization reduces words to their base or dictionary form (lemma).

 Define Stop Words with an example.

Stop words are common words (e.g., "is," "the") often removed in NLP tasks.

 What is Named Entity Recognition (NER)?

NER identifies and classifies entities like names, locations, and dates in text.

 What are N-grams in NLP?

N-grams are continuous sequences of n words in a given text.

 Define Part-of-Speech (POS) tagging.

POS tagging assigns grammatical categories (like noun, verb) to words.

 What is a Language Model?

A language model predicts the probability of word sequences in text.

 What is the purpose of Stemming in NLP?

Stemming reduces words to their root form by removing suffixes.

 Mention any two NLP libraries in Python.

NLTK and spaCy.
QUESTIONS BASED ON PROGRAM:

1. Word Analysis (Frequency and Distribution of Words)

 What is word frequency analysis?

Counting how often each word appears in a text.

 How does your program handle case sensitivity?

By converting all text to lowercase.

 How do you manage punctuation in the text?

By removing or ignoring punctuation marks.

 Which data structure did you use to store word counts? Why?

A dictionary for fast key-based access.

2. Word Generation Using Probabilistic Models

 What is a probabilistic model in the context of word generation?

A model that predicts the next word based on probability.

 How does your program decide which word to generate next?

By selecting words based on their probability distribution.

 What is the difference between bigram and trigram models?

Bigrams consider one previous word, trigrams consider two.

 How does increasing the order of n-grams affect the output?

It improves context but requires more data.

3. Morphology Analysis (Root Words, Prefixes, Suffixes)

 What is morphology in NLP?

The study of word structure, including roots, prefixes, and suffixes.

 How does your program identify root words?

Using stemming or lemmatization techniques.

 What libraries or algorithms did you use for morphological analysis?

NLTK’s stemmer or spaCy’s lemmatizer.

 What is the difference between stemming and lemmatization?

 Stemming chops off affixes; lemmatization finds the dictionary form.

4. Implementing N-Grams

 What is an N-gram?

A sequence of n words from a text.

 How does your program handle the start and end of a sentence?

By adding start and end tokens.

 What is the effect of changing the value of 'n' in N-grams?

Higher n gives better context but increases sparsity.

 How do you handle unseen word sequences?

Using smoothing techniques.

 Can you explain the real-world applications of N-grams?

Text prediction, autocomplete, and speech recognition.

5. N-Grams Smoothing

 Why is smoothing important in N-gram models?

To handle zero probabilities for unseen word sequences.

 What type of smoothing technique did you implement?

Laplace (add-one) smoothing.

 How does smoothing improve model performance?

By assigning small probabilities to unseen events.

 What happens if you don’t apply smoothing?

The model will assign zero probability to unseen sequences.

 How can you evaluate the effectiveness of smoothing techniques?

By measuring perplexity or model accuracy.

6. POS Tagging Using Hidden Markov Model (HMM)

 What is a Hidden Markov Model?

A statistical model where the system has hidden states (like POS tags).

 How does HMM help in POS tagging?

By modeling sequences of tags based on observed words.

  What are the hidden states and observations in your model?

Hidden states are POS tags; observations are words.

 How do you estimate transition and emission probabilities?

Using training data with tagged sentences.

 What are the limitations of HMM for POS tagging?

Difficulty handling long-range dependencies and unknown words.

7. POS Tagging Using Viterbi Decoding

 What is the Viterbi algorithm?

A dynamic programming algorithm for finding the most likely sequence of hidden states.

 Why is Viterbi decoding efficient for POS tagging?

It reduces the computational complexity of sequence prediction.

 How does your program initialize the Viterbi matrix?

By assigning probabilities to the starting states.

 What is the role of backtracking in the Viterbi algorithm?

To trace the optimal sequence of POS tags.

 How does Viterbi decoding handle ambiguous words?

By selecting the tag sequence with the highest probability.

8. Building a POS Tagger

 What approach did you use to build the POS tagger?

A statistical model using HMM and Viterbi decoding.

 How does your program deal with unknown words?

By using smoothing or assigning default probabilities.

 What datasets did you use to train your POS tagger?

Tagged corpora like the Penn Treebank.

 How do you evaluate the accuracy of your POS tagger?

By comparing predicted tags with a labeled test set.

 Can your POS tagger be improved with more data? How?

Yes, more data improves probability estimates and accuracy.

9. Chunking (Grouping Words into Phrases)

 What is chunking in NLP?

Grouping words into meaningful phrases like noun or verb phrases.

 How is chunking different from POS tagging?

POS tagging labels individual words, while chunking groups them.

 What are chunk patterns, and how are they defined?

Regular expressions based on POS tag sequences.

 How does your program recognize noun phrases?

Using patterns like (DT + JJ* + NN+).

 Can chunking be applied to languages other than English?

Yes, with language-specific POS tagging models.

10. Building a Chunker

 What algorithm or library did you use to build the chunker?

NLTK’s RegexpParser for pattern-based chunking.

 How do regular expressions help in chunking?

They define rules for grouping words based on POS tags.

 How do you evaluate the performance of your chunker?

Using precision, recall, and F1 score against annotated data.

 What challenges did you face while building the chunker?

Handling complex sentence structures and ambiguous tags.

 How would you extend the chunker to handle more complex syntactic structures?

By adding more detailed patterns and rules.