Context-based Arabic Word Sense Disambiguation using Short Text Similarity Measure

Word Sense Disambiguation (WSD) is the process of determining which sense of a word is used in a given context. Most of Arabic WSD systems are based generally on the information extracted from the local context of the word to be disambiguated by computing the number of overlapping words between the two concepts definitions. This information is not usually sufficient for a best disambiguation. Because of the short nature of concept definition, we believe that exploiting semantic short text similarity measure can improve the identification process of which sense of a word is used in a context. In this paper, we propose an efficient method for computing the semantic relatedness between senses. To this end, we reintroduce the Web-based Kernel function for measuring the semantic relatedness between concepts to disambiguate an expression versus multiple possible concepts. The proposed method has been tested, evaluated and compared using an Arabic short text categorization system in term of the F1-measure. The obtained results show the interest of our proposition.