Evolution of Semantic Similarity—A Survey

Estimating the semantic similarity between text data is one of challenging and open research problems in field Natural Language Processing (NLP). The versatility natural language makes it difficult to define rule-based methods for determining measures. To address this issue, various have been proposed over years. This survey article traces evolution such beginning from traditional NLP techniques as kernel-based most recent work on transformer-based models, categorizing them based their underlying principles knowledge-based, corpus-based, deep neural network–based methods, hybrid methods. Discussing strengths weaknesses each method, provides a comprehensive view existing systems place new researchers experiment develop innovative ideas issue similarity.