基于情感分析混合模型的用户评论主题分析——以vivo手机为例

为了帮助商家了解用户需求和产品问题,进一步提升产品和服务质量,对vivo手机评论数据进行基于关键词的词云图分析、社会语义网络分析、舆情时间序列可视化分析,挖掘用户关注焦点与手机特征的内在联系和用户情感倾向趋势;然后对评论数据进行基于LDA的主题特征分析,继而提出一种基于Word2vec和SVM、LDA的混合算法模型,挖掘用户正向和负向情感评论的潜在主题,得到不同情感倾向下用户对vivo手机不同方面的反映情况。分析结果表明,基于混合算法的挖掘结果比基于关键词的可视化分析、基于LDA的主题分析更清晰,更具准确性,为商家提供的建议更有意义。

User Comments Subject Analysis Based on Sentiment Analysis Hybrid Model: Taking vivo mobile phone as an example

In order to help merchants understand user needs and product issues, and further improve product and service quality, aiming on on Vivo mobile phone review data,the internal connection between user focus and mobile phone characteristics and the trend of user sentiment tendency are explored through word cloud graph analysis, social semantic network analysis, and public opinion time series visualization analysis of comments data based on keywords . Then, the LDA topic feature analysis has been done. And the hybrid algorithm based on Word2vec, SVM and LDA is put forward to mine the potential topics of the user''s positive and negative emotional comments, and get the user''s reaction to different aspects of the Vivo mobile phone under different emotional tendencies. The analysis results show that the mining results based on the hybrid algorithm are clearer and more accurate than the visual analysis based on keywords and the topic analysis based on LDA, and the suggestions for businesses are more helpful.