基于多个异构社交网络数据分析的用户建模及其应用

With the development and promotion of Internet applications, social networking has become an important part of life and practice. How to conduct user modeling by integrating multiple heterogeneous social networks is an important issue. This project systematically studies the technologies: (1) link users across multiple heterogeneous social networks by a collective Fellegi-Sunter approach; (2) integrate user profiles across multiple heterogeneous social networks by the collaborative filtering; (3) provide the personalized recommendation on multiple heterogeneous social networks by some matrix factorizations. To provide data and platform supports for the related researches of user modeling, we are going to design and implement a prototype system for modeling users across multiple heterogeneous social networks. Conducting the research of this project is helpful to the research of Psychology as well as Social Science. It also has great value in wide applications, such as e-commerce, the management of network environment, online social networking, etc. This project starts from linking users across multiple heterogeneous social networks, it makes an innovation in the technical route. Due to the theoretical results of user linkage across multiple heterogeneous social networks, clarify the application requirements and collected social network data, our research group has a good background and experience on user modeling across multiple heterogeneous social networks. The project is therefore clear in objectives and strong in feasibility.

随着互联网应用的推广和深入，社交网络成为人们生活和实践的重要组成部分。如何综合多个异构社交网络的数据对用户进行建模是一个重要的研究课题。本课题旨在：运用集群Fellegi-Sunter方法关联多个异构社交网络上的用户;利用协同过滤技术融合多个异构社交网络中的用户画像片段;利用矩阵分解的协同过滤技术在多个异构网络中提供相关的个性化推荐;并在此基础上设计并实现一个用户建模的原型系统，为用户建模相关研究提供数据与平台上的支持。这些问题的解决不仅有助于社会学和心理学的研究，而且还有助于电子商务、网络环境管理和在线社交等信息消费行业中的应用。课题以多个异构社交网络上的用户关联作为切入点，在技术路线上具有新意。课题的研究基于课题组在多个异构社交网络上的用户关联的理论成果、厘清的应用需求和收集的网络数据，课题组具有较好的相关研究基础，项目目标明确，可行性较强。

随着互联网应用的推广和深入，社交网络成为人们生活和实践的重要组成部分。如何综合多个异构社交网络的数据对用户进行建模是一个重要的研究课题。本课题旨在：运用集群Fellegi-Sunter方法关联多个异构社交网络上的用户;利用协同过滤技术融合多个异构社交网络中的用户画像片段;利用矩阵分解的协同过滤技术在多个异构网络中提供相关的个性化推荐;并在此基础上设计并实现一个用户建模的原型系统，为用户建模相关研究提供数据与平台上的支持。在本项目资助下，发表论文19篇；申请专利和软件著作权各一项，培养博士生2名、硕士生4名和本科生6名。相关研究成果在腾讯公司实现了成果转化。