Course: B.Tech (Semester V) Subject: Natural Language Processing (NLP)
Subject Code: KAI-052 Time: 3 Hours
Maximum Marks: 100
Instructions: Attempt all sections; assume missing data suitably.
Section A – Short Answer Questions (2 × 10 = 20 Marks)
Answer all questions briefly:
Differentiate between bigram and trigram.
Define the N-grams model.
What are the problems with PCFG (Probabilistic Context-Free Grammar)?
What do you understand by ambiguity?
Give similarities and differences among imitation, synthetic, artificial, fake, and simulated.
Explain Word Sense Disambiguation (WSD).
Explain Articulatory Phonetics.
What are the reasons to use short-time Fourier transform?
List the various issues of a machine translation system.
What is discourse planning?
Section B – Descriptive Questions (10 × 3 = 30 Marks)
Attempt any three:
Explain the need for smoothing and calculate P(Sam | am) using a bigram model with Laplace smoothing for the given corpus:
Discuss the advantages and disadvantages of deep vs shallow parsing. List various types of parsers.
Discuss relations among word senses and knowledge sources in WSD.
Explain speech coding and speech synthesis with pattern matching.
Describe pattern comparison techniques, including Cepstral and Weighted Cepstral distances.
Section C – Long Answer / Analytical Questions (10 × 5 = 50 Marks)
Attempt one part of each question:
Q3.
(a) Write and explain an algorithm for parsing a finite-state transducer, or
(b) Write and explain an algorithm for a simple top-down parser.
Q4.
(a) Given grammar and lexicon, show the final chart for bottom-up chart parser for the sentence “Find the men in suits.”, or
(b) Write a detailed account of the CYK parser.
Q5.
(a) Discuss knowledge sources in Word Sense Disambiguation, or
(b) Compare words eat and find in restriction-based sense disambiguation.
Q6.
(a) Define Articulatory Phonetics and explain production and classification of speech sounds with examples, or
(b) Write regular expressions for:
All alphabetic strings
Lowercase strings ending in “b”
Strings from alphabet {a, b} where each “a” is immediately surrounded by “b”
Explain their use in speech processing.
Q7.
(a) Explain feature extraction and pattern comparison techniques in speech analysis and discuss likelihood and spectral distortions, or
(b) Explain Hidden Markov Model (HMM) with Baum–Welch parameter re-estimation and its implementation issues.
<s> I am Sam </s>
<s> Sam I am </s>
<s> I am Sam </s>
<s> I do not like green eggs and Sam </s>
