Which statement best describes a Markov model?

Which statement best describes a Markov model?

• A. A Markov model predicts the next word based on the entire history of words.
• B. A Markov model predicts the next word based only on a small fixed context.
• C. A Markov model ignores context.
• D. A Markov model relies on semantic similarity.

Answer: (B)

Markov models use the idea that the future depends only on a limited, recent context, not on the entire past. In language tasks this is implemented by a fixed-size window of previous words—the model looks at the last n−1 words to predict the next one (an n-gram approach). This limited memory is the essence of the Markov property and is what makes such models tractable and predictable. So, predicting the next word based only on a small fixed context is the best description. It aligns with the idea of a fixed-order memory rather than using the full history. Predicting from the entire history would violate the Markov assumption, ignoring context misses the probabilistic dependence on past tokens, and relying on semantic similarity reflects different modeling goals than the Markov framework.

Markov models use the idea that the future depends only on a limited, recent context, not on the entire past. In language tasks this is implemented by a fixed-size window of previous words—the model looks at the last n−1 words to predict the next one (an n-gram approach). This limited memory is the essence of the Markov property and is what makes such models tractable and predictable.

So, predicting the next word based only on a small fixed context is the best description. It aligns with the idea of a fixed-order memory rather than using the full history. Predicting from the entire history would violate the Markov assumption, ignoring context misses the probabilistic dependence on past tokens, and relying on semantic similarity reflects different modeling goals than the Markov framework.