A Construction Of Sentiment Topic Model-Based On Non-Parametric Bayesian Methods

With the appearance of microblog, forums, and ecommerce in recent years, a large number of user joined it with passion, the huge number of hot commodity's and hot news comments are generated. It is a great value for researching of products and monitoring of public opinion for government. Therefore, the analysis process of these text information becomes more important and the text sentiment analysis is one of the main core technology. In this paper, we have analyzed the fine-grained emotion analysis, and constructed a sentiment topic model based on non-parametric Bayesian methods.At the beginning, When analysis the traditional sentiment model, it faces two main problems in analyzing user's emotion of product reviews: 1) the lack of fine-grained emotion analysis for product attributes, 2) the number of product attributes should be defined in advance.Then, In order to alleviate problems mentioned above, a fine-grained model for product attributes named User Sentiment Model(USM)was proposed. Firstly, the entities was clustered in product attributes by Hierarchical Dirichlet Processes(HDP) and the number of product attributes could be obtained automatically. Then, the combination of the weight of entities in product attributes, the evaluation phrase of product attributes and sentiment lexicon was considered as prior. Finally, Latent Dirichlet Allocation(LDA) was used to classify the emotion of product attributes.At the last, collected the iphone's reviews on Taobao and Jingdong. The experimental results showed that the model achieved a high accuracy in sentiment classification and the average accuracy rate of sentiment classification is 87%. Compared with the traditional sentiment models, the proposed model obtained higher accuracy on extracting product attributes as well as sentiment classification of evaluation phrases.