基于互动的用户社会行为挖掘

Recently, the growing personalized web applications make a variety of personalized technology emerged, most of these techniques need analyzing users’ network data to discover their interests and behavior laws, and this process is called user behavior analysis. At present, the approaches of user behavior analysis are based on the user's own historical behavior data, which did not consider the impact of social factors on user behavior, such as interactive effects among friends on user behavior. To this end, this project will address the problem of user behavior be influenced by user interact. From basis of user interactive, the short-term effects of user interactive and the long-term effects to research user social behavior based on user interactive. Specially, we will carry out the following researches: 1) in the aspect of user interactive basis, we will study user link behavior. According to the content of the historical message and the current friendships of users in social network, we will design a new hybrid approach by incorporating user interests and user friendships together to recommend new friends for target users; 2) in the aspect of user interactive short-term effects, we will incorporate the friendship of user to assist mining user preferences and will design a new model by integrating user's own behavior and friend interaction;3)in the aspect of user interactive long-term effects, we will study on the influence of users’ interactive and their own factors on the evolution of users’ behavior and interests. The research results of this project have important theoretical value and application prospect in the fields of social network analysis, such as commercial advertising targeting and public opinion on social networks.

日益增长的个性化网络应用促使各种个性化技术应运而生，这些个性化技术需要分析用户网络数据，以发现其兴趣和行为规律，即用户行为分析。当前的用户行为分析主要基于用户自身历史行为数据，未考虑社会因素对用户行为的影响，比如朋友互动等社会因素对用户行为的影响。为此，本项目将针对朋友互动影响用户行为这一问题，从用户行为互动的基础、用户行为互动短期效果、用户行为互动长期效果这三个方面出发，研究基于互动的用户社会行为。具体开展以下研究内容：1）互动基础，研究融合消息内容和已有链接结构的社会链接预测方法；2）互动短期效果，研究基于链接关系的用户兴趣挖掘模型；3）互动长期效果，研究社会互动和用户自身因素对用户行为和兴趣演变。本项目的研究成果对社交网络商业广告推荐、社交网络舆情分析等领域具有重要的理论价值和应用前景。

日益增长的个性化网络应用促使各种个性化技术应运而生，这些个性化技术需要分析用户网络数据，以发现其兴趣和行为规律，即用户行为分析。当前的用户行为分析主要基于用户自身历史行为数据，未考虑社会因素对用户行为的影响，比如朋友互动等社会因素对用户行为的影响。为此，本项目将针对朋友互动影响用户行为这一问题，从用户行为互动的基础、用户行为互动短期效果、用户行为互动长期效果、跨网络用户互动这四个方面出发，研究基于互动的用户社会行为。具体开展以下研究内容：1）互动基础，研究面向好友关系预测的用户嵌入表示方法；2）互动短期效果，研究基于链接关系的用户属性挖掘模型；3）互动长期效果，研究社会互动和用户自身因素对用户行为和兴趣演变。4）探索跨网络用户互动，研究跨网络用户对齐方法。本项目的研究成果对社交网络舆情分析等领域具有重要的理论价值和应用前景。