Text Sentiment Analysis Based on Improved Naive Bayes Algorithm

Aiming at the lack of specific domain corpus in text sentiment polarity analysis, the inaccurate classification accuracy of the naive Bayes algorithm due to the independence assumption and the sparse word vector matrix, a text sentiment analysis method based on the improved naive Bayes algorithm is proposed. Combining machine learning methods with domain sentiment dictionary weighting methods. The improved word frequency inverse file frequency algorithm is used to extract the feature word vector of hotel review text, and the weight of the feature word vector of the domain dictionary after regression test is introduced to weaken the influence of the independence assumption. The singular value decomposition algorithm realizes the dimensionality reduction of the word vector sparse matrix and eliminates redundancy. The remaining features are used to construct a polynomial model of Naive Bayes. The results of simulation research show that this method can effectively improve the effect of text sentiment classification.