社交网络中数字媒体的扩散跟踪及关联建模

项目名称： 社交网络中数字媒体的扩散跟踪及关联建模

项目编号： No.61202241

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 韦世奎

作者单位： 北京交通大学

项目金额： 23万元

中文摘要： 有别于传统媒体数据，从博客、微博及社交网站等社交网络获取的媒体数据及与之相关的社会属性紧密地混合在一起，综合体现了个体和群体的社会行为，构成一种新的媒体形式，称之为社交媒体。一些重大敏感事件往往是由热点社交媒体的快速扩散而引发，其对现实社会产生重大影响，也对国家公共安全造成冲击。跟踪热点社交媒体的扩散路径、并对这些相互交织的社交媒体进行关联建模，实现孤立数据源的信息共享，对整体理解社交媒体内容、发现热点事件、预测事件演变趋势等公共安全应用具有重要意义。本课题面向公共安全这一重大应用，拟从高效的媒体表达入手，以媒体的重复检测为突破口，研究社交媒体的扩散跟踪及关联建模机制，为构建国家公共安全应急信息平台提供基础技术支撑。具体研究内容包括：研究高效的媒体表达方法，建立社交媒体内容的精炼描述；研究可视媒体的重复检测方法，实现重复热点媒体的检测；研究社交媒体关联模型，建立多源孤立媒体信息的语义关联。

中文关键词： 数据表达；模式识别；拷贝检测；关联建模；

英文摘要： Compared with traditional digital media, the data obtained from social network such as blog, microblog and social network sites is closely mixed with some social information, which forms a new media manner and reflects the social behavior of both individuals and groups. We call the new media data as social media. The rapid spread of hot social media frequently triggers some important social events, which greatly impacts the real world and leads to the instability of society. Tracking hot social media and modeling the relationship among them can greatly contribute the public safety oriented applications such as understanding media content, detecting hot events, and predicting the developing trend of events. This project faces the great demand in public safety area and attempts to make some studies from three key aspects: highly efficient representation of social media, the duplication detection, and media relationship modeling. In particular, the main research scheme includes: study highly efficient media representation method, which can build compact and effective representation model; study new duplication detection method, which can detect hot social media; study new media relationship modeling method, which can build semantic relationship for isolated information sources.

英文关键词： Data Representation；Pattern Recognition；Copy Detection；Association Modeling；