Natural language processing question bank 05

List down the features of Probabilistic Context Free Grammars (PCFGs)

Question:

What are the features of Probabilistic Context Free Grammars?

Answer:

• Partial solution for grammar ambiguity. A PCFG gives some idea of different reasonable parses.

• PCFG is not lexicalized since its probability estimates are based purely on structural factors.

• Robustness – the problem of grammatical mistakes, and errors in the real texts can be avoided by assigning low probability for non-reasonable (not probable) sentences. That is, PCFG admits everything with low probability.

• Gives a probabilistic language model for English.

• Predictive power of PCFG is comparatively better than that of HMM.

• An n-gram language model considers some local lexical context while measuring the probability of a sentence, whereas a PCFG does not consider the lexical information. Hence, PCFG is worse model than any n-gram model for English.

• In PCFG, the probability of smaller trees are greater than the larger trees.

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

Related questions:

• Why is it better to compute likelihood and probability values in log space?

• One common way probabilities are assigned to rules in a PCFG is to use a treebank. When will some rules have a probability of 0.0? What is the problem if some rule has a probability of 0.0? How can it be overcome?

• You are building an ngram model of a corpus. Should you stem the words and do the counts or leave them in the surface form? Give pros and cons and include what characteristics of the corpus might influence your decision.

• List down the features of PCFGs

• How many parameters are required to fully define a Hidden Markov Model?  

• Differentiate between generative and discriminative models.