图书情报工作 ›› 2016, Vol. 60 ›› Issue (9): 116-122.DOI: 10.13266/j.issn.0252-3116.2016.09.016

• 知识组织 • 上一篇    下一篇

科研社交网络的推荐系统对比分析

刘先红1,2, 李纲1   

  1. 1. 武汉大学信息资源研究中心 武汉 430072;
    2. 河南科技大学管理学院 洛阳 471023
  • 收稿日期:2016-01-27 修回日期:2016-04-14 出版日期:2016-05-05 发布日期:2016-05-05
  • 作者简介:刘先红(ORCID:0000-0003-1648-8293),博士研究生,E-mail:cbtech@whu.edu.cn;李纲(ORCID:0000-0002-8336-4891),教授,博士,博士生导师。
  • 基金资助:

    本文系国家自然科学基金项目"科研团队动态演化规律研究"(项目编号:71273196)研究成果之一。

Comparative Analysis of Recommender Systems of Research Social Networking Service

Liu Xianhong1,2, Li Gang1   

  1. 1. Center for the Studies of Information Resources of Wuhan University, Wuhan 430072;
    2. Management School of Henan University of Science and Technology, Luoyang 471023
  • Received:2016-01-27 Revised:2016-04-14 Online:2016-05-05 Published:2016-05-05

摘要:

[目的/意义]科研社交网络与大众社交网络一样存在信息过载问题,利用推荐系统向科研人员推送个性化信息是解决该问题的重要手段。通过与国外主流科研社交网络相比较,找出我国科研社交网络的推荐系统存在的问题,进而寻求解决之道。[方法/过程]从推荐项目、推荐策略、冷启动方案、用户偏好学习4个方面,对科研之友、学者网、ResearchGate、Academia这4个科研社交网络的推荐系统进行对比。[结果/结论]我国科研社交网络的推荐系统在上述4个方面都与国外同行存在明显的差距,存在推荐项目较少、推荐策略单一、冷启动效果差、用户偏好学习能力弱等问题。针对这些问题,提出改进建议。

关键词: 科研, 社交网络, 推荐系统

Abstract:

[Purpose/significance] Research social networking service has the same problem of information overload as the popular social networking service. The recommender system is an important measure to solve this problem. Compared with the foreign research social networking service, this paper finds out the problems of the recommender system of China's research social networking service,to provide valuable information to solve such problem.[Method/process] This paper compares the recommender systems of four research social networking services of ResearchGate, Academia, Scholarmate and Scholat, from four aspects of recommending item, recommending strategy, cold start scheme and user preference learning method.[Result/conclusion] It finds that the recommender system of research social networking service of China has a obvious gap compared with foreign counterparts in above aspects. The problems include the fewer recommending items, insufficiency recommending strategies, poor effects of cold start, and weak abilities of user preference learning. Finally, it puts forwards some measures to solve these problems.

Key words: research, social networking services, recommender systems

中图分类号: