Sentiment Analysis Based on Chinese Thinking Modes

Abstract

Sentiment analysis is an important research domain for NLP, and currently it mainly focuses on text context. While our research concentrates on the thinking modes, which influence the formation of language. “Spiral graphic mode”, “concreteness” and “scattered view”, are taken into consideration to assist sentiment analysis and classification in this paper. According to these explicit Chinese modes, a Chinese sentiment expression model (CSE) is proposed, which can effectively improve the accuracy of emotion classification. In order to solve the implicit Chinese sentiment expression, Latent Semantic Analysis (LSA) is applied when the CSE model could not classify the implicit emotions accurately. By comparing with two traditional sentiment analysis methods, experimental results show that the performance of sentiment analysis included the Chinese thinking mode factors is significantly better than which not included.