AKTU Natural Language Processing (KCS072) Question Paper 2024-25

Note: Attempt all Sections. In case of any missing data; choose suitably.

Section A

Q1. Attempt all questions in brief · 2 × 10 = 20 marks

1. How does context influence error detection? 1 K 1 ,
2. A Hidden Markov Model (HMM) is used for PoS tagging. Explain the 1 K 1 , backward algorithms used to compute tag probabilities. K
3. Provide an example of unification of feature structures for agreement in 2 K 1 , number and gender.
4. Discuss the limitations of CFGs in modeling natural language syntax. 2 K ,
5. Compare and contrast dictionary-based and distributional methods for word similarity measurement.
6. Define selectional restrictions.
7. How are speech sounds classified? 4 K 1 ,
8. How do vocal tract shape and size affect the spectrum of speech sounds? 4 K ,
9. Demonstrate the Viterbi algorithm. 5 K 3 ,
10. Compare and contrast LPC and PLP coefficients for speech feature 5 K 3 , extraction. K

Section B

Q2. Attempt any three of the following · 10 × 3 = 20 marks

1. Consider the regular expression (ab)*c. Draw the corresponding finite- 1 K 1 , state automaton and explain how it recognizes the language.
2. Given the sentence "The dog saw the man with the telescope," illustrate 2 K 1 , the ambiguity in parsing using dependency grammar. Propose a K resolution.
3. Discuss how syntax-driven semantic analysis works. Create a semantic attachment for the sentence: "John gave Mary a book."
4. Compare filter-bank and LPC methods in speech feature extraction. 4 K 1 , Provide numerical examples where possible. K
5. What are likelihood distortions in speech recognition? Provide examples 5 K 3 , and their perceptual impact.

Section C

Q3

Attempt any one part of the following · 10 × 1 = 10 marks

1. A word processor uses a minimum edit distance algorithm to suggest 1 K 1 , corrections for misspelled words. If the word "intention" is misspelled K as "execution," calculate the minimum edit distance and outline the alignment steps.
2. Compare and contrast interpolation and backoff smoothing techniques. 1 K 1 , How are these applied to n-gram models?

Q4

Attempt any one part of the following · 10 × 1 = 10 marks

1. Discuss the concept of treebanks in NLP. How do they facilitate training 2 K 1 , syntactic parsers? Illustrate with an example.
2. Explain the CYK parsing algorithm with a worked example of parsing 2 K , the sentence "He saw a cat." using a given CFG.

Q5

Attempt any one part of the following · 10 × 1 = 10 marks

1. Given a sentence with multiple possible word senses (e.g., "bank"), outline how Word Sense Disambiguation (WSD) is performed using supervised learning.
2. Implement a bootstrapping method for WSD using a small set of seed 3 K words. Illustrate with examples.

Q6

Attempt any one part of the following · 10 × 1 = 10 marks

1. Explain the mathematical basis of the log-spectral distance measure. 4 K 1 , Compute it for two spectral frames with given power spectra.
2. Derive and explain the LPC coefficients for a given speech frame. 4 K ,

Q7

Attempt any one part of the following · 10 × 1 = 10 marks

1. Discuss spectral distortion measures in speech analysis. Calculate the 5 K 3 , cepstral distance for given cepstral coefficients.
2. Discuss the role of HMMs in speech recognition. Explain the forward 5 K , and backward procedures with an example.

Short on time? See the most repeated AKTU Natural Language Processing questions across all years.

Other years

More AKTU Natural Language Processing papers on LearnAIFast.

This paper is reproduced for study purposes only. Copyright belongs to AKTU. Source: the question paper bank of ABES Institute of Technology, an AKTU-affiliated college. If you own the copyright and want it removed, email learnwithskillnest@gmail.com and we will take it down.