🎲 Probabilistic parsing §

= a parsing strategy to find the most likely parse for ambiguous grammars using 🎲 Probability Theory

How? §

• Augment context free grammars with the 🎲 Probability for each rule
• Calculate the overall probability P(T) of a parse T as the product of all the rules r probabilities:
  • $P(T)={\prod }_{n\in T}p(r(n))$

Problems §

• Independence assumption
• Structural & lexical dependencies