(1&30) What are the main two components of NLP.
List any three basic and three advanced
text preprocessing techniques
(2&31) What is the most probable next word predicted by the model for the following word
sequence using Bi-gram model
<S> Henry? (2M)
Given Corpus:
<S> Henry likes football </S>
<S> I do play football </S>
<S> Do you like football </S>
<S> I like sports </S>
<S> Henry and I play sports </S>
<S> Do I play football </S>
<S> I do like Henry </S>
3) Apply Hidden Markov Model-Viterbi optimization for the following Labeled Corpus
<S> Martin(N) Justin (N) can (M) watch (V) will (N) <E>
<S> Spot (N) will (M) watch (V) Martin (N) <E>
<S> Will (M) Justin (N) Spot (V) Martin (N) <E>
<S> Martin (N) will (M) pat (V) Spot (N) <E>
Sentence: 'Justin will spot Will'
4) Categorize and explain the approaches of POS tagging also state the challenges of POS
tagging.
5) Consider the following corpus.
<s> I am Sam</s>
<s> Sam I am </s>
<s> Sam I like </s>
<s> Sam I do like </s>
<s> Do I like Sam</s>
i)List all the possible bi-grams with their probabilities.
ii)Predict the next word for: <s> Do?
6) Differentiate Top-down and Bottom-Up approach of parsing. Construct the parsing tree for
the given sentence using Top-down parser 'Sentence: The dog chases the cat'
Given Grammar:
S → NP VP
NP → Det N
VP → V NP
Det → “the”
N → “dog” | “cat”
V → “chased”
(7&33) Explain HMM in detail. Explain the limitation during this process and state the name
of the algorithm to overcome it.
8) Illustrate all the semantic and syntactic constraints on coherence with examples.
(9&32) List the word embedding techniques and explain Word2Vec word embedding
technique
10) Differentiate between Syntactic analysis and Lexical analysis
(11&17) Define ambiguity in NLP and List any two ambiguities with example
12) State the reasons why Natural language processing is difficult for computers.
13) Differentiate between Syntactic ambiguity and Lexical Ambiguity.
(14&29) Consider the following corpus of three sentences
1. "AI solves problems."
2. "Machines learn AI."
3. "AI improves machines."
Represent each sentence as a binary vector using one-hot encoding
15) Explain the term NLP and list any five applications of NLP
18) Explain the architecture of NLG Systems
(19&20) What is Machine Translation & list Its Challenges
21) Explain Characteristics of Indian Languages
22) What is WordNet? explain Structure and applications of WordNet
23) How Information Retrieval Systems (IR Systems) works? draw architecture diagram,
explain Components of an IR System and Challenges of IR System
24) Explain POS Tagger
25) Explain IR based question answering system
26) What is entity linking in the context of Question Answering systems, and why is it crucial
for accurate information retrieval?
27) Describe the Dialogue State Architecture. How does it help manage the flow of
conversation in frame-based dialogue systems?
28) What are the key metrics used for evaluating dialogue systems and chatbots? Discuss the
challenges involved in evaluating human dialogue
34) Explain any one following Technique for Feature Extraction from Text
1. Bag of Words (BoW)
2. Term Frequency - Inverse Document Frequency (TF-IDF)
3. One-Hot Encoding
4. N-Grams
35) Write a note on any one of the following Tokenization
-Stemming
-POS Tagging
-Shallow Parsing
-Coreference Resolution
-Lemmatization
-Chunking
-Word Embedding
-Parsing
36) What are Recurrent Neural Networks (RNNs), and how do they handle sequential data?
Explain one major limitation of RNNs when used for long sequences.
37) How does an LSTM network overcome the vanishing gradient problem in RNNs? Describe
the role of the forget gate, input gate, and output gate in an LSTM cell.
38) Compare GRU with LSTM in terms of architecture and performance. Why might one
choose GRU over LSTM in certain NLP tasks?
39) What is the attention mechanism in neural networks, and how does it improve
performance in sequence-to-sequence models? Provide an example use case in NLP.
40) What is the Transformer model in NLP, and how does it differ from RNN-based models?
Explain the role of self-attention and positional encoding in the Transformer architecture.
