基于机器学习的企业产品评论数据的情感分析研究

Emotional Analysis of Enterprise Product Review Data Based on Machine Learning

简要分析了中小企业风险控制和管理问题,借助海量网页数据,利用LDA+SVM组合算法,对行业新闻进行智能分类,利用流式数据对企业产品评论数据进行情感分析研究。鉴于流式数据的稀疏性问题,提出基于信息字节N元语法、信息量、评论情感极性等进行特征扩展的方法,并结合SVM算法,进行观点挖掘。实验结果表明分类器性能有着一定程度的提升。研究结果能够帮助企业管理者全面快速了解市场和把握消费者对产品的态度,及时规避风险。

This paper makes a brief analysis on the risk control and management of SMEs by analyzing the massive webpage data.The industry news is classified intelligently by using LDA+SVM combination algorithm,and the emotional analysis of enterprise product review data is carried out by using streaming data.In view of the sparsity of streaming data,a feature extension method based on N-ary grammar of information bytes,information quantity and comment emotional polarity is proposed,and the viewpoint mining method is combined with SVM algorithm.Experimental results show that the classifier performance is improved to a certain extent.The research results can help business managers understand the market comprehensively and quickly,grasp consumers'attitudes towards products,and avoid risks in time.