近年来,随着网络社交媒体的兴起与不断发展,社交媒体在网络用户的生活中的影响越来越大,大规模的网络用户积极参与到社交媒体中来,形成了十分活跃的虚拟社会，引起了政府、业界和学界多层次的关注。从学界来看，可分为两个主要方面，一是考察以往在社会网络分析中得到的经典结论在网络社交媒体上体现的形式和程度，二是考察在网络社交媒体上是否存在新的基础性原理。而其中用户行为模式的分析、理解、与挖掘成为一个非常关键的研究问题。本课题定位在社交媒体中的用户兴趣建模,进而来更好地理解社会媒体,以及其内 部信息传播的一些基本模式和规律。本课题提出的假说是：尽管用户兴趣纷繁复杂且动态变化，但层次式结构能对它们进行有效地表达。我们希望通过探索相关理论并建立一套具有层次性的动态用户兴趣模型来支持这个假说，并希望所建立的用户兴趣模型能改进一系列针对社交媒体的分析和挖掘工作。研究中，我们将特别考虑结合我国的社交媒体的特点。

With the rapid growth of online social networks, they have significantly changed the way that users live. A large number of Web users actively are engaged in a diverse range of activities in social networks and interact with one another to form a virtual society. Therefore, there is a emerging need to construct deeply research on analyzing, understanding and mining on online social media. Current research focus fall on two questions: 1) whether the existing principles are still fit for online social media or not, and 2) whether there would be new principles as guildlines for further expoloration in social media. One of the most key challenges is to develop a series of powerful methodologies to understand the user interests and behaviors. .This proposal aims to better mine the basic principles or rules in online social networks via effectively learning users' interests. Online content is generated in a dramatically fast way and the semantics embedded in it can be very complicated, which make it difficult for simple user interest models to discover the underlying knowledge. We hypothesize that user interest models with a hierarchical structure can better capture the basic patterns in spite of the complex nature of Web social media. So we propose to build dynamic hierarchical user interest models to adapt to these characteristics in online social media, which is expected to improve the downstream related applications. Besides the possible scientific publications, we will derive a comprehensive analyisis of online social media for our country, which can be beneficial to the development of Web research in China.