AL-504-B Natural Language Processing - Jun 2025 Question Paper

Under each question, How to answer lists the points an answer should cover, outlined by CampusPrep from the syllabus. Check the details against your textbook.

1. 7 Marks
   aDefine formal grammar and discuss its role in language modeling.
   How to answer

   • Define formal grammar as tuple (N, Sigma, P, S) and types
   • Explain productions, derivation and language generated
   • Role in language modeling: syntax specification, parsing, constraints
   • Relation to statistical models and grammar-based LM
2. 7 Marks
   bExplain the need for Transducers in morphological parsing.
   How to answer

   • Define morphological parsing and transducer/FST concept
   • Need: two-level mapping surface to lexical form
   • Handling inflection, derivation, compounding efficiently
   • Bidirectional analysis and generation benefit
3. 7 Marks
   aWrite the regular expression pattern for the following [email protected] and 1111 1111 1111 1111.
   How to answer

   • Give regex for email pattern [email protected] form
   • Explain components: local part, @, domain, TLD
   • Give regex for 16-digit grouped credit-card-like pattern
   • Show handling of dots and spaces
4. 7 Marks
   bWhat are some common types of smoothing techniques used in NLP, and how do they differ from each other?
   How to answer

   • Define smoothing and zero-probability problem in n-grams
   • Describe Laplace/Add-one, Add-k, Good-Turing
   • Describe backoff and interpolation, Kneser-Ney
   • Compare discounting, context use and performance
5. 7 Marks
   aDifferentiate key components of Hidden Markov and Maximum Entropy models.
   How to answer

   • Define HMM generative components: states, transitions, emissions
   • Define MaxEnt discriminative components: features, weights, constraints
   • Contrast generative vs discriminative, independence assumptions
   • Contrast training and use in sequence labeling
6. 7 Marks
   bWrite short notes on PoS tagging types and issues in word analysis.
   How to answer

   • Define PoS tagging and tagsets
   • Types: rule-based, stochastic, transformation-based
   • Issues: ambiguity, unknown words, sparse data
   • Word analysis challenges in morphologically rich languages
7. 7 Marks
   aExplain how dependency grammar differs from phrase-structure grammar.
   How to answer

   • Define phrase-structure grammar: constituency, rewrite rules
   • Define dependency grammar: head-dependent relations
   • Contrast representation: tree vs dependency links
   • Example sentence parsed both ways
8. 7 Marks
   bDescribe disambiguation techniques and challenges in parsing.
   How to answer

   • Define syntactic ambiguity types: lexical, structural, PP-attachment
   • Techniques: probabilistic PCFG, chart pruning, supertagging
   • Challenges: combinatorial explosion, garden path, long dependencies
   • Evaluation and context use
9. 7 Marks
   aCheck the acceptance of string $s=\text{\"ababab\"}$ using the CYK Algorithm. for the given grammar $$S \rightarrow AB / BC$$ $$A \rightarrow BA / a$$ $$B \rightarrow CC / b$$ $$C \rightarrow AB / a$$
   How to answer

   • State grammar in CNF and CYK table setup for length 6
   • Fill diagonal with A/C for a and B for b, then combine
   • Check if S in cell [0,6] to decide acceptance
   • Conclude accepted/rejected with parse
10. 7 Marks
    bIllustrate Word similarity using a thesaurus and distributional methods.
    How to answer

    • Define word similarity and thesaurus-based measure (WordNet)
    • Explain distributional hypothesis and co-occurrence vectors
    • Compare: manual knowledge vs corpus-based, coverage
    • Example cosine similarity calculation
11. 7 Marks
    aBriefly explain different Bootstrapping methods in NLP.
    How to answer

    • Define bootstrapping/semi-supervised learning need
    • Explain self-training, co-training, Yarowsky algorithm
    • Explain Hearst patterns / snowball for relation extraction
    • Advantages and semantic drift issue
12. 7 Marks
    bExplain supervised Word Sense Disambiguation methods with suitable examples.
    How to answer

    • Define WSD and supervised formulation as classification
    • Methods: Naive Bayes, decision lists, SVM, neural classifiers
    • Features: context window, collocations, PoS
    • Example with word 'bank' and training data
13. 7 Marks
    aHow can natural language processing create intelligent tutoring systems? Discuss with applications.
    How to answer

    • Define intelligent tutoring systems components
    • NLP roles: NLU, dialogue, essay scoring, feedback
    • Examples: question answering, language learning tutors
    • Benefits and challenges
14. 7 Marks
    bDescribe the functioning of machine translation and provide examples of its real-world applications.
    How to answer

    • Define machine translation and Vauquois triangle levels
    • Types: rule-based, statistical, neural MT with encoder-decoder
    • Pipeline: analysis, transfer, generation / training and decoding
    • Real-world apps: Google Translate, localization, cross-lingual IR
15. 14 Marks
    Explain any two of the following:

    1. a)Tokenization
    2. b)Unification of feature structures
    3. c)Syntax-Driven Semantic analysis
    4. d)Man-Machine interfaces
    How to answer

    • Select any two topics for detailed notes
    • Give definition, key steps/formalism and example for each
    • Mention uses in NLP pipeline
    • Mention limitations if relevant