Segmentation thématique des textes Anglais

Abstract

The Topical Text Segmentation (TTS) is an important task in many natural language processing (NLP) applications, such as information retrieval (IR), automatic text summarization and question answering. It consists of dividing the texts into segments, each segment corresponds to a different topic. The methods used are mainly probabilistic (statistical), which sometimes leads to the failure of the TTS system due to the semantic ambiguity and the impossibility to identify the semantic relations between the words. Our task is to develop an english thematic text segmenter based on a conceptuel representation. That is why we have integrated the ontology WordNet English as a semantic resource.