基于领域词典的民宿评论情感分析  

Sentiment Analysis of Homestay Comments Based on Domain Dictionary

作　　者：杨鑫 杨云帆 焦维 朱东霖 郑绍阳 袁中玉 杨秀璋 罗子江[1] YANG Xin;YANG Yun-fan;JIAO Wei;ZHU Dong-lin;ZHENG Shao-yang;YUAN Zhong-yu;YANG Xiu-zhang;LUO Zi-jiang(Information College,Guizhou University of Finance and Economics,Guiyang 550025,China)

机构地区：[1]贵州财经大学信息学院,贵阳550025

出　　处：《科学技术与工程》2020年第7期2794-2800,共7页Science Technology and Engineering

基　　金：国家自然科学基金(11664005);贵州省科技计划(黔科合基础[2019]1041号);贵州财经大学2019年度在校学生科研资助项目(2019ZXSY77);贵州省研究生教育创新计划项目(黔教合YJSCXJH[2019]066)。

摘　　要：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.

关 键 词：民宿评论 情感词典 情感分析 SO-PMI LDA 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]