Virtual Labs

What is the main problem that smoothing solves in N-gram language models?

a: It reduces the size of the vocabulary Explanation

Explanation

b: It assigns non-zero probabilities to unseen N-grams Explanation

Explanation

c: It increases the speed of computation Explanation

Explanation

d: It removes stopwords from the corpus Explanation

Explanation

In Add-One Smoothing, what is added to each bigram count?

a: Zero Explanation

Explanation

b: One Explanation

Explanation

c: The vocabulary size Explanation

Explanation

d: The total number of bigrams Explanation

Explanation

Given a vocabulary size of 5 and a bigram count C('the', 'cat') = 2, and C('the') = 4, what is the Add-One smoothed probability P('cat'|'the')?

a: 2/4 Explanation

Explanation

b: 3/9 Explanation

Explanation

c: 2/9 Explanation

Explanation

d: 3/4 Explanation

Explanation

Which smoothing technique is considered a simple baseline for N-gram models?

a: Good-Turing Smoothing Explanation

Explanation

b: Kneser-Ney Smoothing Explanation

Explanation

c: Add-One (Laplace) Smoothing Explanation

Explanation

d: Backoff Smoothing Explanation

Explanation

Why is Add-One Smoothing not always preferred for real-world language modeling?

a: It is too complex to implement Explanation

Explanation

b: It can overestimate the probability of unseen N-grams Explanation

Explanation

c: It requires removing stopwords Explanation

Explanation

d: It only works for unigrams Explanation

Explanation

Which of the following is the correct formula for Add-One smoothed bigram probability?

a: P(wi|wi-1) = (C(wi-1, wi) + 1) / (C(wi-1) + V) Explanation

Explanation

b: P(wi|wi-1) = C(wi-1, wi) / (C(wi-1) + 1) Explanation

Explanation

c: P(wi|wi-1) = (C(wi-1, wi) - 1) / (C(wi-1) + V) Explanation

Explanation

d: P(wi|wi-1) = (C(wi-1, wi) + V) / (C(wi-1) + 1) Explanation

Explanation

If a bigram ('she', 'likes') appears 0 times in the corpus, C('she') = 2, and V = 5, what is the Add-One smoothed probability P('likes'|'she')?

a: 0/2 Explanation

Explanation

b: 1/7 Explanation

Explanation

c: 1/5 Explanation

Explanation

d: 2/7 Explanation

Explanation

Which of the following tasks would most benefit from N-gram smoothing?

a: Predicting the next word in a sentence Explanation

Explanation

b: Counting the number of sentences Explanation

Explanation

c: Sorting words alphabetically Explanation

Explanation

d: Finding the longest word Explanation

Explanation

What happens to the probability of seen N-grams after smoothing is applied?

a: It always increases Explanation

Explanation

b: It always decreases Explanation

Explanation

c: It may decrease slightly to allow for unseen N-grams Explanation

Explanation

d: It becomes zero Explanation

Explanation

N-Grams Smoothing