Textual Entailment for Arabic Language based on Lexical and Semantic Matching

Mariam Khader, Arafat Awajan, Akram Alkouz
2016 International Journal of Computing and Information Sciences  
Textual Entailment is one of the recent natural language processing challenges, where the meaning of an expression "Hypothesis" could be entailed by the meaning of another expression "Text". In comparison with English Language, the Arabic language has more challenges in determining entailment judgment, due to lexical and syntactic ambiguity. The proposed work in this paper has adopted a lexical analysis technique of Textual Entailment to study the suitability of this technique for Arabic
more » ... e. In addition a semantic matching approach was added in order to enhance the precision of the proposed entailment system. The lexical analysis is based on calculating word overlap and bigram extraction and matching. The semantic matching has been incorporated with word overlap to increase the accuracy of words matching. The system has been evaluated by measuring precision and recall, those two metrics are the main evaluation measures used in Recognizing Textual Entailmen challenge number 2 to evaluate participated systems. The system has achieved precision of 68%, 58% for both Entails and NotENtails respectively with overall recall of 61%.
doi:10.21700/ijcis.2016.109 fatcat:4qtoznozyfc37lkjj4fd7jeck4