Machine Translation Evaluation using Textual Entailment for Arabic

Textual entailment is a generic task aiming to capture semantic inference between two text fragments. It has been applied so far to improve many natural language applications such as Question Answering and Information Extraction for the English language. This work aims to exploit results obtained for Textual Entailment in machine translation evaluation to Arabic. We show that the output of a system built for textual entailment recognition for Arabic can be used to serve as a metric to evaluate machine translation into Arabic. Despite the simplicity of the concepts and techniques., this metric has realized competitive results in comparison to state-of-the-art tools in terms of correlation with human judgments.