Estudio estadístico del número de reglas resultantes al transformar una gramática libre de contexto a la forma normal de chomsky

Abstract

It is well known that any context-free grammar can be transformed to the Chomsky normal form so that the languages generated by each one are equivalent. A grammar in Chomsky Normal Form (CNF), has some advantages: Their derivation trees are binary, simplest rules and so on. So it is always desirable to work with a grammar in CNF in applications that require them. There is an algorithm that can transform a context-free grammar to one CNF grammar, however the number of rules generated after the transformation depends on the initial grammar and other circumstances. In this work we analyze from the experimental and statistical point of view the relationship between the number of initial rules and the number of resulting rules after transforming. This allows you to plan the amount of computational resources needed in case of dealing with grammars of some complexity.