基于领域词典的民宿评论情感分析

Sentiment Analysis of Homestay Comments Based on Domain Dictionary

通过基于领域词典的情感分析法,从用户生成的内容中更为准确地分析其情感状态,为民宿业提供一种新的研究视角。以贵阳民宿评论为研究样本,采用SO-PMI算法完成领域词典的构建,并借助LDA主题模型和可视化技术对用户评论进行情感分析。研究发现,构建的领域词典相较基础情感词典而言,性能上得到提升,尤其在负面评论方面,准确率、召回率上分别提升了17%和16%。同时结合LDA主题挖掘,详尽分析民宿评论中的正负面主题并分析其内在原因,这能为民宿管理者做出更好的决策提供数据支持和理论支撑。

The emotional state of user-generated content more accurately was analyzed through the sentiment analysis method based on domain dictionary,and a new research perspective for the homestay industry was provided.Guiyang homestay comments were taken as research samples,the SO-PMI algorithm was used to complete the construction of the domain dictionary,and the user comments were analyzed by LDA theme model and visualization technology.It was found that the domain dictionary constructed has improved performance compared to the basic sentiment dictionary.Especially in the aspect of negative comments,the accuracy rate and recall rate have also increased by 17%and 16%respectively.Combined with the LDA theme mining,the positive and negative themes in homestay reviews and their internal causes were analyzed in detail,which can provide data support and theoretical support for the administrator of the homestay to make better decisions.