高质量用户生成内容主题分布特征研究  

Study of Topic Distribution Features of User Generated Comments of High Quality

作　　者：阮光册 夏磊[2] Ruan Guangce;Xia Lei(Vacuity of Economics and Management, East China Normal University;Shanghai Library)

机构地区：[1]华东师范大学经济与管理学部 [2]上海图书馆

出　　处：《图书馆杂志》2018年第4期95-101,共7页Library Journal

摘　　要：随着社会化媒体的发展,识别海量高质量的用户生成内容在电子商务等领域具有很多的实际意义。本文将主题模型应用到高质量用户生成内容的识别,从语义层面挖掘高质量用户生成内容所具有的特征。本文首先通过对社交平台用户评论特征的分析,构建了基于LDA的用户生成内容主题模型,对主题模型计算结果,采用TFIDF识别高概率主题词的权重,探讨高质量用户生成内容的主题分布特征。最后通过亚马逊网上书店的网络用户评论进行试验,分析了高质量用户评论的主题分布特征。With the development of the social media, it is important to identify the user generated comments of high quality. In this paper, the author used the topic model to identify the high quality users＇ comments and to mine the topic feature. The author first put forward the topic model in high quality users＇ comments, then calculated the topic model and discussed the topic distribution features. Based on the analysis of the users＇ comments on the social platforms, the author established LDA subjects model, and used the users＇ comments in Amazon bookstore for testing. Finally, the author drew a conclusion on the topic distribution feature on high quality users＇ comments,

关 键 词：用户生成内容 高质量 TFIDF 主题模型 

分 类 号：G252[文化科学—图书馆学]