Virtual Labs

POS Tagging - Hidden Markov Model

What is the primary purpose of Part-of-Speech (POS) tagging in natural language processing?

a: To count the number of words in a sentence Explanation

Explanation

b: To assign grammatical categories like noun, verb, adjective to each word Explanation

Explanation

c: To translate words between languages Explanation

Explanation

d: To identify the meaning of words Explanation

Explanation

In a Hidden Markov Model, what are the 'hidden' states?

a: The words in the sentence Explanation

Explanation

b: The POS tags that we want to predict Explanation

Explanation

c: The probabilities in the model Explanation

Explanation

d: The training data Explanation

Explanation

What does the word 'Markov' refer to in Hidden Markov Models?

a: The assumption that future states depend only on the current state Explanation

Explanation

b: The name of the algorithm used Explanation

Explanation

c: The type of probability distribution used Explanation

Explanation

d: The language being processed Explanation

Explanation

What are emission probabilities in an HMM for POS tagging?

a: The probability of one POS tag following another Explanation

Explanation

b: The probability of a word being generated given a POS tag Explanation

Explanation

c: The probability of starting a sentence with a particular POS tag Explanation

Explanation

d: The probability of ending a sentence with a particular POS tag Explanation

Explanation

Why is context important in POS tagging?

a: Because words always have only one possible POS tag Explanation

Explanation

b: Because many words can have multiple POS tags depending on context Explanation

Explanation

c: Because it makes the algorithm faster Explanation

Explanation

d: Because it reduces the vocabulary size Explanation

Explanation

In the calculation P(tag₂|tag₁) = count(tag₁, tag₂) / count(tag₁), what does this represent?

a: Emission probability Explanation

Explanation

b: Transition probability Explanation

Explanation

c: Initial probability Explanation

Explanation

d: Final probability Explanation

Explanation

What is the main advantage of using statistical methods like HMMs over hand-crafted rules for POS tagging?

a: They are always 100% accurate Explanation

Explanation

b: They can automatically learn patterns from data and handle ambiguity Explanation

Explanation

c: They require less computational resources Explanation

Explanation

d: They work only with English language Explanation

Explanation

In the Viterbi algorithm, what does dynamic programming help achieve?

a: It reduces the search space by storing optimal subproblem solutions Explanation

Explanation

b: It increases the accuracy of the model Explanation

Explanation

c: It handles unknown words better Explanation

Explanation

d: It improves the training speed Explanation

Explanation

What is a major limitation of first-order HMMs for POS tagging?

a: They cannot handle any ambiguous words Explanation

Explanation

b: They assume the next tag depends only on the current tag, not on longer contexts Explanation

Explanation

c: They require too much training data Explanation

Explanation

d: They only work with short sentences Explanation

Explanation