ISSN 1000-1239 CN 11-1777/TP

计算机研究与发展 ›› 2019, Vol. 56 ›› Issue (9): 1927-1938.doi: 10.7544/issn1000-1239.2019.20180723

• 信息处理 • 上一篇    下一篇


左笑晨1, 窦志成2, 黄真1, 卢淑祺1, 文继荣2   

  1. 1(中国人民大学信息学院 北京 100872); 2(大数据管理与分析方法研究北京市重点实验室(中国人民大学) 北京 100872) (
  • 出版日期: 2019-09-10
  • 基金资助: 

Product Category Mining Associated with Weibo Hot Topics

Zuo Xiaochen1, Dou Zhicheng2, Huang Zhen1, Lu Shuqi1, Wen Jirong2   

  1. 1(School of Information, Renmin University of China, Beijing 100872); 2(Beijing Key Laboratory of Big Data Management and Analysis Methods (Renmin University of China), Beijing 100872)
  • Online: 2019-09-10
  • Supported by: 
    This work was supported by the National Key Research and Development Plan of China (2018YFC0830703), the National Natural Science Foundation of China (61872370), and the Fundamental Research Funds for the Central Universities (2112018391).

摘要: 微博是目前人们广泛使用的在线分享和交流的社交媒体平台之一.某些被广泛关注的话题因为在微博中被大量网友转发、评论和搜索而形成微博热门话题,而这些热门话题的广泛传播则可能进一步刺激和推动用户的线下行为.作为其中的典型代表,某些微博热门话题可能会刺激电商平台中和该话题相关的商品的热销.提前挖掘出与微博热门话题相关联的商品品类,可帮助电商平台和卖家提前做好商品运维以及库存的调配,提高用户搜索的购物转化率,带来相应商品销量的提升.提出了一种微博热门话题所关联的潜在购物品类的挖掘方法.首先构建商品知识图谱,然后采用多种深度网络模型对商品品类的关联知识图谱信息与微博话题内容进行文本匹配,识别出每个热门话题和商品品类的关联强度.实验表明,该方法能够有效识别出热门话题和购物品类的关联关系,大部分的微博热门话题都可以关联到电商平台中至少一个商品品类.

关键词: 知识图谱, 文本匹配, 微博热点, 实体识别, 深度学习

Abstract: Weibo is one of the widely used social media platforms for online sharing and communication. Some widely-received topics have been formed into Weibo hot topics by being forwarded, reviewed, and searched by a large number of users in Weibo. And the widespread dissemination of these hot topics may further stimulate and promote users offline behaviors. As a typical representative of it, some hot topics on Weibo may stimulate sales of products related to the topics under the e-commerce platform. Mining out the relevant product categories of Weibos hot topics in advance can help e-commerce platforms and sellers to do a good job of commodity operation and inventory deployment as well as promote the search conversion rate of users and bring about an increase in the sales of corresponding products. This paper proposes a method of mining potential shopping categories associated with hot topics of Weibo. First, the method builds a product knowledge map, and then uses a variety of in-depth network models to perform textual matching between the information of the associated knowledge of product categories and the content of the Weibo topics. The strength of association of each hot topic and product category is identified. Experiments show that the method can effectively identify the relationship between hot topics and shopping categories, and most of the hot topics of Weibo can be associated with at least one product category in the e-commerce platform.

Key words: knowledge graph, textual match, Weibo hotspot, entity recognition, deep learning