An hybrid method for the Arabic queries disambiguation to improve the relevance calculation in the IRS
In this article, we introduced a new approach to facilitate the calculation of relevance in information research systems in Arabic langage. Our method is to remove the morphosemantic ambiguity due to agglutination, and lack of vocalization of the Arabic words. To do, we have proposed to transform words to semantic gene. The latter represent an accurate realization of the meaning of the word. They contain the type, the context, the definition and vocalized shape of all possible cases may be taken in the Arabic word. In our approach we consider all the possible meanings of the terms by applying a morphosemantic variation based on a recursive algorithm. Obtained variants are filtering by using of the sentence context, the user profile and the synthesis of an Arabic phrase rules. The result is a semantically coherent text ready to be used by an information search system.
Keywords: semantic gene; Arabic disambiguation; TALN; Information research
Download Full-Text
ABOUT THE AUTHORS
Adil Enaanai
Centre d\'études doctorales ST2I ENSIAS Equipe SIME
Aziz Sdigui Goukkali
Centre d\'études doctorales ST2I ENSIAS Equipe SIME
El Habib Benlahmer
Centre d\'études doctorales ST2I ENSIAS Equipe SIM
Adil Enaanai
Centre d\'études doctorales ST2I ENSIAS Equipe SIME
Aziz Sdigui Goukkali
Centre d\'études doctorales ST2I ENSIAS Equipe SIME
El Habib Benlahmer
Centre d\'études doctorales ST2I ENSIAS Equipe SIM