AL-504-B Natural Language Processing - Nov 2023 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
   aDescribe the role of regular expressions in NLP and provide examples of how they are used in language processing tasks?
   How to answer

   • Define regular expressions and pattern matching formalism
   • Role in tokenization, text normalization and information extraction
   • Examples: email/phone/date matching, token splitting, search/replace
   • NLP tasks: preprocessing, named entity recognition, text cleaning
2. 7 Marks
   bExplain the key components of a Grammarians Language Model (LM) and how it functions in NLP?
   How to answer

   • Define Language Model (LM) and its purpose
   • Key components: vocabulary, probabilities, N-grams/neural parameters
   • Training and inference functioning in NLP
   • Applications: prediction, generation, MT, speech recognition
3. 7 Marks
   aExplore the challenges associated with detecting and Correcting Spelling Errors in natural language text.
   How to answer

   • Types of spelling errors: non-word vs real-word
   • Challenges: ambiguity, context dependence, morphology, proper nouns
   • Detection and correction approaches: edit distance, noisy channel
   • Trade-off between precision, recall and efficiency
4. 7 Marks
   bWhat is smoothing in the context of N-gram models and why is it necessary? Provide example to show how smoothing works?
   How to answer

   • Define smoothing and zero-probability problem
   • Why necessary: unseen N-grams, sparse data
   • Example with Laplace/Add-one smoothing calculation
   • Effect on probability distribution
5. 7 Marks
   aWhat is transformation-based tagging and how does it contribute to improving Part-of-Speech tagging accuracy?
   How to answer

   • Define transformation-based tagging (Brill tagger)
   • Learning transformations from errors
   • How iterative rules improve accuracy over baseline
   • Example of lexical and contextual rules
6. 7 Marks
   bExplain the principles of rule-based Part-of-Speech tagging.
   How to answer

   • Define rule-based POS tagging
   • Principles: lexicon, morphological and contextual rules
   • Disambiguation using hand-crafted rules
   • Strengths and limitations vs statistical methods
7. 7 Marks
   aHow do CFGs capture the syntactic structure of sentences, and what are the key components of a CFG?
   How to answer

   • Define CFG and its 4-tuple: terminals, non-terminals, productions, start symbol
   • How productions capture phrase structure and hierarchy
   • Example derivation / parse tree for a sentence
8. 7 Marks
   bExplain the process of syntactic parsing.
   How to answer

   • Define syntactic parsing and its goal
   • Steps: tokenization, POS tagging, grammar application, tree building
   • Top-down vs bottom-up parsing approaches
   • Output parse tree and handling ambiguity
9. 7 Marks
   aWhat is the process involved in constructing treebanks, and how do they play a role in both the development and assessment of syntactic parsers?
   How to answer

   • Define treebanks with examples like Penn Treebank
   • Construction steps: collection, annotation, parsing, validation
   • Role in training statistical parsers
   • Role in evaluation and benchmarking
10. 7 Marks
    bWhat is Word Sense Disambiguation in NLP?
    How to answer

    • Define Word Sense Disambiguation (WSD)
    • Why needed: polysemy and homonymy
    • Brief approaches: dictionary, supervised, unsupervised
    • Applications in MT, IR, QA
11. 7 Marks
    aExplain the concept of compositional semantics and how it contributes to understanding the meaning of complex expressions?
    How to answer

    • Define compositional semantics and Frege principle
    • How word meanings combine via syntactic rules
    • Example of phrase/sentence meaning construction
    • Contribution to handling complex and novel expressions
12. 7 Marks
    bExplain how speech recognition works, and how NLP enhances this technology?
    How to answer

    • Outline speech recognition pipeline: acoustic and language models
    • Role of NLP: language modeling, decoding, post-processing
    • Handling ambiguity and context
    • Examples of NLP-enhanced accuracy
13. 7 Marks
    aHow does NLP make work processors smarter and more user-friendly?
    How to answer

    • NLP features in word processors: grammar check, autocomplete, style
    • How language models and POS tagging enable them
    • User-friendly benefits: error reduction, productivity
    • Examples: MS Word, Google Docs smart features
14. 7 Marks
    bHow is NLP used in commercial applications to improve user experience and interaction?
    How to answer

    • Commercial uses: chatbots, search, sentiment, recommendation
    • Improvement via personalization and natural interaction
    • Examples: virtual assistants, customer support
    • Impact on user experience and engagement
15. 14 Marks
    Write short notes on following:

    1. a)Dictionary and Thesaurus
    2. b)First order logic
    3. c)Tokenization
    How to answer

    • Dictionary: definitions, POS, pronunciation; Thesaurus: synonyms/antonyms
    • First order logic: predicates, quantifiers, inference
    • Tokenization: splitting text, challenges, types
    • Role of each in NLP pipeline