19ECS443 NATURAL LANGUAGE PROCESSING

2022-23 Odd Semester

SAMPLE QUESTIONS

MODULE-1

1. What is the main objective of Natural Language Processing (NLP)? Name any two applications of
NLP. (2m) (L1)
2. Explain the working of HAL 9000 computer in understanding human language. (5m) (L2)
3. Why the sentence “I made her duck” is ambiguous? (5m) (L1)
4. Define ‘regular expression’. Match the following regular expressions with suitable strings.
(Note down the matching pairs) (5m) (L1)
(a) /!/ (w) “look up ^ now”
(b) /[wW]oodchuck/ (x) “plenty of 7 to 5”
(c) /[0-9]/ (y) “Oh my God!”
(d) /e^/ (z) “Woodchuck”
5. Define Kleene * operator and give an example. (2m) (L1)
6. What is the difference between Kleene * and Kleene + operators? (2m) (L1)
7. Illustrate any two Anchors in regular expressions with examples. (2m) (L2)
8. How do you specify the two strings “puppy” and “puppies” using a single regular expression?
(2m) (L1)
9. Demonstrate (i) Inflected form of words and (ii) Code switching with examples. (5m) (L2)
10. Illustrate ‘word tokenization’ with examples. (5m) (L2)
11. Explain Byte-Pair Encoding process for word tokenization by writing the algorithm and applying
it on the following corpus: (10m) (L2)

5 low_ Column 1 indicates

2 lowest_ the frequency of each
6 newer_ word in the corpus (l
3 wider_ o w _ appears 5
2 new_ times,….)
12. Summarize (i) Word normalization and (ii) Lemmatization with suitable examples. (5m)(L2)
13. Illustrate the ‘sentence segmentation’ step in text processing. (5m) (L2)
14. Compare the calculation of Minimum edit distance between the strings ‘INTENTION’ and
‘EXECUTION’ using ordinary method and Levenshtein distance. (5m) (L2)
15. Outline the Minimum edit distance algorithm. Compute the minimum edit distance for the
strings ‘INTENTION’ and ‘EXECUTION’ for any five cells in the corresponding n x n matrix. (10m)
(L2)
16. Why is it necessary to predict upcoming words in a sentence? (2m) (L1)
17. Define a bigram and trigram. Give examples. (2m) (L1)
18. What is Markov assumption? (2m) (L1)
19. Consider the following mini corpus:
<s> I am Sam </s>
<s> Sam I am </s>
<s> I do not like green eggs and ham </s>
Infer probabilities for any five bigrams. (5m) (L2)

1
20. How do you compute the probability of the sentence “I want English food” given the following
probability values? (2m) (L1)
P(I | <s>) = 0.25 P(want | I) = 0.33
P(English | want) = 0.0011 P(food | English) = 0.5
P(</s> | food) = 0.68
21. What are the two methods used for evaluating a language model? (2m) (L1)
22. Define perplexity. (2m) (L1)
23. Explain zero-probability bigrams with an example. (2m) (L2)
24. Explain smoothing and its two techniques Laplace smoothing and Add-k smoothing. (10m) (L2)
25. Explain the smoothing techniques: (i) Backoff, (ii) Interpolation and (iii) Katz backoff. (6m) (L2)

MODULE-2

1. Illustrate the eight parts of speech in English with an example for each. (4m) (L2)
2. Explain (i) Closed classes and Open classes, (ii) Count nouns and Mass nouns and (iii) the four
types of Adverbs with examples. (6m) (L2)
3. Explain (i) Particle, (ii) Determiner (iii) Auxiliary verb and (iv) Politeness markers with examples.
(4m) (L2)
4. Define any ten tags from the Penn Treebank tagset (provide tag name, description and an
example). (5m) (L1)
5. Label each word in the following sentences with proper tags using Penn Treebank tagset. (i) The
grand jury commented and (ii) There are 70 children there. (4m) (L1)
6. Find one tagging error in each of the following sentences that are tagged with the Penn
Treebank tagset: (i) I/PRP need/VBP a/DT flight/NN from/IN Atlanta/NN and (ii) Does/VBZ
this/DT flight/NN serve/VB dinner/NNS. (4m) (L1)
7. Define any ten tags from the Brown corpus tagset (provide tag name, description and an
example). (5m) (L1)
8. What are the two types of tagging algorithms? (2m) (L1)
9. Demonstrate HMM part-of-speech tagging. Illustrate with the help of the sentence ‘Secretariat
is expected to race tomorrow’. (10m) (L2)
10. Demonstrate transformation-based tagging. (5m) (L2)
11. Explain the evaluation of error using 10-fold crossvalidation and analyzing the error using a
confusion matrix in POS tagging. (4m) (L2)
12. Outline (i) tag indeterminacy, (ii) tokenization and (iii) unknown words. (6m) (L2)
13. How do you tag the sentences (i) It is a nice night and (ii) I like to watch French movies using
Penn Treebank tagset? (4m) (L1)

MODULE-3

1. What is syntactic constituency? (2m) (L1)

2. What are the two evidences we can show for ‘words form constituents’? (2m) (L1)
3. Define a context-free grammar (CFG). (2m) (L1)
4. What are the two symbols used in context-free grammars (CFGs)? Give examples. (2m) (L1)
5. Show the formal definition of a context-free grammar? (2m) (L1)
6. Explain ‘structural ambiguity’ through an example that shows two parse trees for an ambiguous
sentence. (4m) (L2)

2
7. What are the two types of structural ambiguity? Give examples. (2m) (L1)
8. Apply the CKY parsing algorithm on the sentence ‘the flight includes a meal’ by considering the
following grammar with explanation of conversion to Chomsky normal form (CNF): (10m) (L3)
S  NP VP Det  the N  meal
NP  Det N Det  a N  flight
VP  V NP V  includes
9. Apply the Earley parsing algorithm on the sentence ‘Book that flight’ by considering the
following grammar: (10m) (L3)
Det  that | this | a | the
Noun  book | flight | meal | money
Verb  book | include | prefer
S  NP VP
S  VP
NP  Det Nominal
Nominal  Noun
VP  Verb
VP  Verb NP
10. Explain Probabilistic Context-free Grammars (PCFGs) by considering the sentence ‘astronomers
saw stars with ears’ and the following grammar: (8m) (L2)
S  NP VP (1.0) NP  NP PP (0.4)
PP  P NP (1.0) NP  astronomers (0.1)
VP  V NP (0.7) NP  ears (0.18)
VP  VP PP (0.3) NP  saw (0.04)
P  with (1.0) NP  stars (0.18)
V  saw (1.0) NP  telescopes (0.1)

MODULE-4

1. What do you expect from a ‘model of word meaning’? (4m) (L1)

2. Demonstrate Lemmas and Synonymy. (8m) (L2)
3. What is Principle of contrast? (2m) (L1)
4. Compare Word similarity and Word relatedness. (6m) (L2)
5. Illustrate (i) Semantic Frames and Roles and (ii) Connotations. (8m) (L2)
6. What is vector semantics? (4m) (L1)
7. Build a term-document matrix by considering four terms and four documents. (5m) (L3)
8. Explain word similarity through term-term matrix with an example. (5m) (L2)
9. Infer the formula for Cosine similarity measure. Use the following data to decide which word
(cherry or digital) is closer to ‘information’. (10m) (L2)
pie data computer
cherry 442 8 2
digital 5 1683 1670
information 5 3982 3325
10. Explain the process of creating a tf-idf representation of some sample text. (8m) (L2)
11. Explain the working of Skip-gram with Negative Sampling algorithm (word2vec) by showing the
necessary formulas. (8m) (L2)
12. Model the learning of embeddings by skip-gram. Use the training data
... lemon, a [tablespoon of apricot jam, a] pinch ...
with ‘apricot’ as the target word for your explanation. (10m) (L3)

3
13. Illustrate any two methods to visualize embeddings. (2m) (L2)
14. Explain semantic properties of embeddings. (8m) (L2)

****

MODULE-5

1. Definition of discourse model (2m)

2. Coreference with respect to referring expressions (4m)
3. Definition of Coreference resolution (2m)
4. Various types of referring expressions. (8m)
5. Classification of entities informationally. (5m)
6. Four types of structures that are not counted as mentions in coreference tasks. (8m)
7. Any five types of linguistic properties of the coreference relation. (10m)
8. Define coherence and discourse. (2m)
9. The three ways in which real discourses are locally coherent. (6m)
10. Any two examples of global coherence. (2m)
11. What is word sense and WordNet? (2m)
12. What is the role of dictionaries and thesauruses in understanding word senses? (4m)
13. Define zeugma and give an example. (2m)
14. Demonstrate word sense disambiguation. (8m)

****