MS-DS Master of Data science FREE Master of Data Science Natural Language Processing Questions and Answers 3

Question 1

What distinguishes a Transformer architecture from traditional RNN-based models in NLP?

Accepted Answer

It processes all positions in parallel using self-attention instead of sequential recurrence

Answer

It uses convolutional filters instead of attention

Answer

It requires pre-trained word embeddings to function

Answer

It can only handle fixed-length input sequences

Question 2

In sentiment analysis, what challenge does sarcasm detection present for standard classifiers?

Accepted Answer

The literal meaning of words contradicts the intended sentiment

Answer

Sarcastic sentences are always shorter than sincere ones

Answer

Sarcasm only occurs in spoken language, not text

Answer

Standard tokenizers cannot parse sarcastic sentences

Question 3

What is the role of positional encoding in Transformer-based NLP models?

Accepted Answer

To inject information about token order since self-attention has no inherent notion of position

Answer

To reduce the dimensionality of input embeddings

Answer

To normalize attention weights across heads

Answer

To convert subword tokens back into full words

Question 4

Which tokenization strategy splits rare words into smaller subword units to handle out-of-vocabulary terms?

Accepted Answer

Byte-Pair Encoding (BPE)

Answer

Whitespace tokenization

Answer

Character-level one-hot encoding

Answer

Lemmatization

Question 5

What does perplexity measure when evaluating a language model?

Accepted Answer

How well the model predicts a held-out test set, with lower values indicating better performance

Answer

The number of unique tokens the model can generate

Answer

The ratio of training loss to validation loss

Answer

The average sentence length the model produces

Question 6

In topic modeling with Latent Dirichlet Allocation (LDA), what does each document's topic distribution represent?

Accepted Answer

The mixture of latent topics that generated the words in that document

Answer

The TF-IDF scores for each word in the document

Answer

The probability of the document belonging to a single fixed category

Answer

The syntactic parse tree of the document