Discarding Impossible Events from Statistical Language Models
Résumé
This paper describes a method for detectingvimpossible bigrams using a vocabulary of V elements. The idea is to discard all the ungrammatical events to expect an improvement of the language model. We extract the impossible bigrams by using automatic rules. The biclass associations which are ungrammatical are detected and all the corresponding bigrams are analyzed and set as possible or impossible events, we also decided to manage for each of the retrieved rule an exception list.
Loading...