图书情报工作 ›› 2021, Vol. 65 ›› Issue (10): 79-89.DOI: 10.13266/j.issn.0252-3116.2021.10.009

• 知识组织 • 上一篇    下一篇

网络社交平台中社群标签生成研究

蒋武轩, 易明, 熊回香, 童兆莉   

  1. 华中师范大学信息管理学院 武汉 430079
  • 收稿日期:2020-12-09 修回日期:2021-03-18 出版日期:2021-05-20 发布日期:2021-06-02
  • 作者简介:蒋武轩(ORCID:0000-0001-9621-4318),博士研究生,E-mail:jiangchair@mails.ccnu.edu.cn;易明(ORCID:0000-0002-4864-6025),教授,博士生导师;熊回香(ORCID:0000-0001-9956-3396),教授,博士生导师;童兆莉(ORCID:0000-0003-1621-4356),博士研究生。
  • 基金资助:
    本文系国家社会科学基金年度项目"融合知识图谱和深度学习的在线学术资源挖据与推荐研究"(项目编号:19BTQ005)研究成果之一。

Research on the Generation of Community Tags in Network Social Platform

Jiang Wuxuan, Yi Ming, Xiong Huixiang, Tong Zhaoli   

  1. School of Information Management, Central China Normal University, Wuhan 430079
  • Received:2020-12-09 Revised:2021-03-18 Online:2021-05-20 Published:2021-06-02

摘要: [目的/意义] 基于网络社交平台中社群话题及用户兴趣挖掘而生成的社群标签,能够提高社群定义的及时性与准确性,解决用户信息获取、网络社群选择的困难。[方法/过程] 通过对网络社群的深入分析,发现社群特征可根据社群话题及用户兴趣予以表征。首先,利用主题提取BTM模型对网络社群话题进行主题模型训练,从而得到网络社群话题预标签;其次,根据社群成员兴趣标签网络中不同类型的重要节点指标,利用TOPSIS多指标综合评价方法挖掘成员整体兴趣,从而得到网络社群成员兴趣预标签。综合两者结果生成社群标签并进行优化,且以"豆瓣小组"为例进行实证。[结果/结论] 基于社群话题及成员兴趣的社群标签生成模型能够准确地挖掘主要兴趣及近期关注点,社群整体的标签生成有利于网络用户兴趣群体的选择。

关键词: 社群标签, 标签生成, BTM, TOPSIS

Abstract: [Purpose/significance] Community tags generated based on the mining of community topics and users' interests in network social platforms can improve the timeliness and accuracy of the definition of community, and solve the difficulties of user information acquisition and network community selection. [Method/process] Through in-depth analysis of the network community, it was determined that the community features can be represented according to the community topics and users' interests. Firstly, the BTM model of topic extraction was used to train the topic model of network social topics, and the pre-label of network social topics was obtained. Then, based on the different important node indexes of community members' interest tag network, the TOPSIS multi-index comprehensive evaluation method was used to mine the overall interest of members, so as to obtain the interest pre-label of members of the network community. After combining the two results, the community tag was generated and optimized. And this paper took "Douban Group" as an example for demonstration. [Result/conclusion] The community tag generation model based on community topics and members' interests can accurately mine the main interests and recent concerns. Tag generation of the community as a whole is conducive to the selection of interest groups of network users.

Key words: community labels, tag generation, BTM, TOPSIS

中图分类号: