Meaning representation for automatic indexing of Arabic texts
The aim of indexing is to identify the words that represent the main idea of a paragraph or a specific text, in the framework of the representation of the meaning in an automatic treatment (NLP) of the Arabic; we propose a model based on conceptual vectors. These vectors try to represent the whole of ideas contained in textual segment (word, expression, texts …). This model lean on modern linguistic conception the semantic field theory. By basing itself on the semantic relations (synonymy, homonymy...) between the words, we use these fields for the construction of semantic field data base and of a vectorial space then we calculate the meaning of textual segments in the semantic fields. Finally we use this model for indexing the text.
Keywords: Semantic modeling, NLP, Automatic indexing, Arabic language, Semantic field, Conceptual vector
Download Full-Text
ABOUT THE AUTHORS
Bakhouche Abdelali
for preparation a PhD in natural language processing
Yamina Tlili-Guiassa
Laboratory LRI, University Badji Mokhtar Annaba, Algeria
Bakhouche Abdelali
for preparation a PhD in natural language processing
Yamina Tlili-Guiassa
Laboratory LRI, University Badji Mokhtar Annaba, Algeria