Advanced Database Management System - Tutorials and Notes: Solved Questions and Answers in NLP 04

Search Engine

Please visit, subscribe and share 10 Minutes Lectures in Computer Science

Wednesday, 27 May 2020

Solved Questions and Answers in NLP 04

Natural language processing quiz questions with answers, NLP true false interview questions, NLP quiz questions for competitive exams



NLP TRUE/FALSE Quiz Questions - SET 04

1. While computing the language model probabilities in log format, we multiply the log probabilities to derive the probability of the given model.
(a) TRUE                                          (b) FALSE

View Answer

Answer: (b) FALSE
Adding in log space is same as multiplying in linear space. Later we can find the exp of added log probabilities to get back the probability of the given model as follows;
p1 * p2 * p3 = exp(log p1 + log p2 + log p3)
Example:
Let us suppose, p1 = 0.025, p2 = 0.033, and p2 = 0.05.
p1 * p2 * p3 = 0.025 * 0.033 * 0.05 = 0.00004125 which is very small.
Instead of the above calculation, we can perform the following;
Exp(log p1 + log p2 + log p3) = exp(-1.6 + (-1.48) + (-1.3)) = exp(-4.38) = 10-4.38 = 0.00004169.

2. To measure the quality of a language model, we use the metric intrinsic evaluation on the training data set.
(a) TRUE                                          (b) FALSE

View Answer

Answer: (b) FALSE
Intrinsic evaluation is a metric to measure the quality of a language model independent of any application. And for this, we need test data set.
Also, we do not need external applications to evaluate the performance as in the case of extrinsic evaluation.

3. Since the language model is meant to assign non-zero probability to unseen strings of words, a mandatory desirable property is Smoothing.
(a) TRUE                                          (b) FALSE

View Answer

Answer: (a) TRUE
Unknown words (out-of-vocabulary words) usually cause the problem in a language model. The probability value of unknown words will be zero, and due to multiplication of probabilities, the entire probability will become zero. To avoid zero probabilities, we need to do smoothing.
Some smoothing techniques are Laplace smoothing, Add-k smoothing, Interpolated Kneser-Ney smoothing etc.

4. The production rules of Context Free Grammar (CFG) should be in Chomsky Normal Form (CNF).
(a) TRUE                                          (b) FALSE

View Answer

Answer: (b) FALSE
The grammar of CFG is need not be in Chomsky normal form.
Production rules of CFG can have a combination of non-terminal and terminal symbols on the right hand side. In CNF, a rule can have only one terminal on its RHS (no other symbols) or two non-terminals on it RHS.


5. In backoff smoothing technique, we use the bigrams if the evidence for trigram is insufficient.
(a) TRUE                                          (b) FALSE

View Answer

Answer: (a) TRUE
In backoff, we use the trigram if the evidence is sufficient, otherwise we use the bigram, otherwise the unigram. In other words, we only “back off” to a lower-order n-gram if we have zero evidence for a higher-order interpolation n-gram.
It is a non-linear method. The estimate for an n-gram is allowed to back off through progressively shorter histories. The most detailed model that can provide sufficiently reliable information about the current context is used.

*************************




Top interview questions in NLP

NLP quiz questions with answers explained

Bigram and trigram language models

Online NLP quiz with solutions





No comments:

Post a comment

Featured Content

Multiple choice questions in Natural Language Processing Home

MCQ in Natural Language Processing, Quiz questions with answers in NLP, Top interview questions in NLP with answers Multiple Choice Que...

All time most popular contents