图书情报工作 ›› 2017, Vol. 61 ›› Issue (7): 137-144.DOI: 10.13266/j.issn.0252-3116.2017.07.019

• 综述述评 • 上一篇    

异构信息网络融合方法研究综述

田鹏伟1,2, 张娴1, 胡正银1, 董坤1,2, 许海云1   

  1. 1. 中国科学院成都文献情报中心 成都 610041;
    2. 中国科学院大学 北京 100190
  • 收稿日期:2017-01-06 修回日期:2017-03-20 出版日期:2017-04-05 发布日期:2017-04-05
  • 作者简介:田鹏伟(ORCID:0000-0001-6784-3617),硕士研究生;张娴(ORCID:0000-0002-6297-1190),副研究员,硕士生导师,通讯作者,E-mail:zhangx@clas.ac.cn;胡正银(0000-0002-5699-9891),副研究员,博士;董坤(ORCID:0000-0001-8455-9),博士研究生;许海云(ORCID:0000-0002-7453-3331),副研究员。
  • 基金资助:
    本文系中国科学院科技服务网络计划(STS计划)项目"中国科学院知识产权信息服务"(项目编号:KFJ-SW-STS-190)与广东省省级科技计划项目"再生医学和组织工程知识集成服务技术研发与应用"(项目编号:2016A040403098)研究成果之一。

Review of Studies on Heterogeneous Information Network Fusion Methods

Tian Pengwei1,2, Zhang Xian1, Hu Zhengyin1, Dong Kun1,2, Xu Haiyun1   

  1. 1. Chengdu Library and Information Center, Chinese Academy of Sciences, Chengdu 610041;
    2. University of Chinese Academy of Sciences, Beijing 100190
  • Received:2017-01-06 Revised:2017-03-20 Online:2017-04-05 Published:2017-04-05

摘要: [目的/意义] 异构信息网络融合对于异构信息网络本身及其相关应用意义重大。综述异构信息网络融合方法,并进行客观的分析和评价,以期为进一步研究提供新的思路。[方法/过程] 在对异构信息网络及其相关概念进行辨析的基础上,对异构信息网络融合方法进行调研、分析与归纳,评述该领域的研究现状,提出未来可能的研究方向。[结果/结论] 异构信息网络融合方法分为基于元路径提取、多重关系网络及超图/超网络建模等五种类型,具体方法各有优势与局限;当前异构信息网络融合研究尚处起步阶段,研究方法有待丰富;基于元路径提取的融合方法显现不足;基于异构信息网络融合的应用型研究需进一步开拓。

关键词: 异构信息网络, 网络融合, 元路径, 超图, 超网络, 聚类, 分类

Abstract: [Purpose/significance] The study of heterogeneous information networks (HINs) fusion is of great significance for heterogeneous information network analysis and their applications. In this paper, some typical methods are reviewed in order to provide new ideas for further research. [Method/process] Based on the comparison of the concepts between HINs and other networks, this paper investigated, analyzed and summarized the research methods of HINs fusion, reviewed the current research in this field, and put forward the future research directions possibly. [Result/conclusion] The methods of HINs fusion are divided into five types: meta-path extraction, multi-relational network and hyper-graph/super-network modeling, and so on. and these specific methods have their own advantages and limitations. At present, HINs fusion is still at the initial stage. The research methods and applications based on HINs fusion needs further exploration.

Key words: heterogeneous information network, network fusion, meta-path, hyper-graph, super-network, clustering, classification

中图分类号: