Probability distribution on the parse trees.

Abstract

Paper describes probability distribution on the parse trees of natural language by using Bayesian networks. First parts of the paper describes probabilistic context-free grammar and parse trees. In the second part of the paper, Bayesian network was modelled and joint probability distribution on their vertex. On these theoretical ideas, in the third part, we describe our model tested on Database of grammatical sentences of Croatian language (http:/ /infoz.ffzg.hr / tepes /). At the end was presented a backward procedure and evaluation of our results.