Review of approaches for paraphrase identification

The article is devoted to a review of approaches solving the problem identifying paraphrases. This problem's relevance and use in tasks such as plagiarism detection, text simplification, information search are described. Several classes solutions were considered. first approach based on manual rules - it uses manually selected features fundamental properties second lexical similarity various databases ontologies. Machine learning-based also presented this paper describe different architectures that can be used identify last considered deep learning modern models transformers.