Sentence Similarity - A State of Art Approaches

This is the authors' abstract. We don't add key points for this paper.

This paper draws up a summary of the different approaches to sentence similarity methods. Sentence similarity has a important place in different applications of natural language processing. Plagiarism check, question answering, information retrieval, text summarization, text classification are some of the natural language processes where sentence similarity finds its applications. The existing works on text similarity have been discussed by approaching through 3 methods: String-based, Knowledge-based, and Corpus-based similarity. Relatedness between short texts is measured by each approach based on a specific perspective. Moreover, the introduction of datasets that are mostly used as benchmarks for the evaluation of techniques in this field provides an absolute view on this issue. Better results are obtained when approaches that combine more than one perspective are used.