Automatic Summarization Of Arabic Texts

Résumé: This paper presents an automatic summarizer for Arabic texts. The main goal of a summarizer is to produce a condensed version of the content of an input text for various users. The approach exploits mainly the expressions of the language present in the text denoting sentences' relevance according to the author's point of view regardless of the domain. The summarizer selects the most relevant sentences based on linguistic knowledge in the form of linguistic patterns representing language expressions and word lists. Language expressions express the language knowledge and are independent from any specific domain. The paper presents the linguistic acquisition process which feeds the linguistic knowledge base, the architecture of the system including three modules, and the system implementation.

Publié dans la revue: AL-Lisaniyyat