Abstract:

Paraphrase identification allows computing the degree of semantic equivalence between source and suspect documents. This represents a challenge in many Natural Language Processing (NLP) applications (e.g. text summarization, information retrieval, text categorization, etc.). In this context, we addressed the problems of sentence meaning and word order for detecting paraphrase in Arabic Language. Often distributed word representation models have gained promising results for analogy reasoning and similarity analysis.