Python Natural Language Processing
上QQ阅读APP看书,第一时间看更新

Approach to handle syntactic ambiguity

To handle this ambiguity, we need to use statistical approaches and get the most likelihood ratio. We need to take co-occurrences between the verb and the preposition in one hand, and preposition and the noun on other hand, and then calculate the log-likelihood ratio by using following equation:

Figure 3.14: Log-likelihood ratio

Here, p(p/v) is the probability of seeing a PP with preposition p and after verb v.

And p(p/n) is the probability of seeing a PP with preposition p after noun n.

If F(v,p,n) < 0, then we need to attach the preposition to the noun, and if F(v,p,n) >0, then we need to attach preposition to the verb. We will see the implementation in Chapter 5, Feature Engineering and NLP Algorithms.