A Novel Discrimination Structure for Assessing Text Semantic Similarity

<p>Discrimination of semantic textual similarity refers to comparing the between two or more entities (including words, short texts and documents) through certain strategies obtain a specific quantitative value. Traditional research put experience into calculation original text content, using matching degree distance characters words as yardstick judge whether pairs are similar. However, there still some problems be solved in following aspects: key points sentence meaning word semantics, which play important role expression natural language, not well integrated discrimination, interactive features fully utilized. To solve above problems, this paper proposes novel discrimination structure based on Siamese Network model idea matching. In structure, we introduce information realize extraction interaction feature information, then vector representation by BiLSTM. The experimental results showed that accuracy proposed is higher than basic models.</p> <p> </p>