Ar-SLoTE: A Recognizing Textual Entailment Tool for Arabic Question/Answering Systems

Abstract

Recognizing the relation of entailment between sentences is an important and common part of linguistic communication. The recognizing textual entailment task has been proposed as a solution for this problem. In this paper, we present an Arabic Recognizing Textual Entailment Tool called Ar-SLoTE «Arabic Semantic Logical Textual Entailment Tool». The proposed tool is composed of five modules: pretreatment, linguistic analysis, first-order logic representation, features extraction and entailment decision modules. It extracts the logical representations of the hypothesis/text pairs in order to extract valuable and informative features namely, predicates-arguments overlap, semantic similarity and named entity matching. Ar-SLoTE is destined especially to Arabic factual question/answering systems and the attained result is very encouraging.