Short text similarity measurement methods: a review

Short text similarity measurement methods play an important role in many applications within natural language processing. This paper reviews the research literature on short (STS) method with aim to (i) classify and give a broad overview of existing techniques; (ii) find out its strengths weaknesses terms domain independence, requirement semantic knowledge, corpus training data, ability identify meaning, word order polysemy; (iii) knowledge resource that can be utilized for STS methods. Furthermore, our study also considers various issues such as difference between sources corpora similarity. Although there are few review papers this area, they focus mostly only one/two techniques. do not cover recent research. To best is comprehensive systematic topic. The findings follows: It identified four eight resources external classified into general-purpose domain-specific. techniques string-based, corpus-based, knowledge-based hybrid-based. Moreover, expert researchers utilize benchmark well reference limitations current identifies open considered feasible opportunities future directions.