摘要 为了解决SUV车型外观海量评论文本隐藏信息挖掘分析问题,给出了一个基于LDA主题模型的数据挖掘方法。通过LDA主题模型识别出评论文本中潜藏的主题信息,计算感兴趣的文本主题和文本涵盖的主题比例。经过情感信息抽取、情感信息分类和情感分析建模等步骤,实现对文本评论的倾向性判断和隐藏信息挖掘,得到SUV车型的用户情感倾向分析结果,并挖掘出特定SUV车型外观的优缺点。为汽车企业在外观设计用户评价方面提供一种可供借鉴的解决方案。 A data mining method based on LDA model was proposed for extracting hidden information from massive data of SUV model appearance reviews.The hidden themes in the review texts were identified by applying LDA,and the proportion of the themes of interest was calculated.The sentiment analysisis conducted such as sentiment extraction,sentiment classification etc.,the emotional orientations were determined and the data mining of review texts were accomplished.Finally the analysis results of users’sentiment tendency on SUV models are obtained.The features of some specific SUV models are excavated as well.This paper provides a methodological guidance for automobile enterprises to find a solution in product appearance design based on user evaluation.
出处 《汽车工程学报》 2021年第2期93-101,共9页 Chinese Journal of Automotive Engineering
基金 江苏省高校自然科学研究重大项目“非标定光度立体视觉三维重建技术及其应用研究”(19KJA510006)。
关键词 情感分析 评论文本 LDA主题模型 汽车外观 sentiment analysis review text latent dirichlet allocation model automobile appearance
