A Study of Chinese Semantic Analysis Based on Enhanced Lexicon

Sentiment analysis using natural language processing and machine learning techniques can determine the sentiment or emotional tone expressed in large amounts of text. However, due to the presence of Chinese polysemous words, current sentiment analysis lacks effective analysis for various specific domains. As a result, sentiment analysis in Chinese faces challenges. To address this issue, this paper presents a novel domain-specific sentiment analysis system. The system combines the advantages of two major sentiment analysis methods: lexicon-based sentiment analysis and machine learning analysis. First, the Naive Bayesian classification algorithm is used to extract words, accurately discriminate their emotional polarity, and form an enhanced lexicon suitable for a specific domain. Then, three commonly used machine learning classifiers are used to compare the effects of semantic analysis before and after using the enhanced lexicon. Experimental statistics show that the sentiment orientation accuracy of the Chinese financial commentary sentiment analysis system proposed in this paper can be improved by about 5%.