机器学习情感分析方法改进研究

Improvement Study of the Sentiment Analysis Method in Machine Learning

近年来,随着社交网络的迅速发展,舆情监督成为国内外研究的热点。由于网民参与评论的渠道多种多样,因此,需要对网络用户进行网络言论监测。舆情分析的基础是情感分析技术,然而,现存的情感分析技术存在着不足,准确率难以得到保证。基于对情感分析方法进行改进,以及进一步提高分析结果准确率的研究目标,通过采用知网情感词典并对其合并扩展的情感分析方法和基于机器学习的SVM和KNN情感分析方法,在比较了基于情感词典的情感分析方法以及基于机器学习的情感分析方法的优缺点后,提出了用情感词典和机器学习方法相结合的方式进行情感分析。实验表明,结合情感词典以及SVM和KNN加权方式提高情感分类的准确率较之前提升了近5个百分点,准确率明显提高。

In recent years,with the rapid development of social networks,public opinion supervision has become a hot topic at home and abroad.Because there are various channels for netizens to participate in comments,it is necessary to monitor online users'comments.The basis of public opinion analysis is emotional analysis technology.However,the existing emotional analysis technology is insufficient,and the accuracy is difficult to be guaranteed.Based on the improvement of the emotion analysis method and the research goal of further improving the accuracy of the analysis results,the paper suggests combining emotion dictionary and machine learning for sentiment analysis by adopting the emotion analysis method of HowNet emotion dictionary and merging and expanding it and the emotion analysis method of SVM and KNN based on machine learning,and after comparing the advantages and disadvantages of the emotion analysis method based on emotion dictionary and the emotion analysis method based on machine learning.The experiment shows that the accuracy of emotion classification improved by combining emotion dictionary,SVM and KNN weighting method is nearly 5 percentage points higher than before,and the accuracy is significantly improved.