项目名称: 面向实时复杂数据的信息扩散模型和分析算法研究
项目编号: No.61502169
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 自动化技术、计算机技术
项目作者: 姚俊杰
作者单位: 华东师范大学
项目金额: 21万元
中文摘要: 随着社交网络和移动智能设备的快速普及,互联网上发布的信息不仅规模巨大,同时其时效性也显著加快,传播范围明显扩大。尤为突出的是,微博、点评、互动和位置记录等新场景可以实现更实时的内容发布和反馈。与此同时,其关联的附加特征也使得数据处理任务越发复杂。信息扩散研究结合了时序信息和社交网络领域,建模内容变迁和用户行为,近期获得了大量关注,但是针对这些新的特征的研究较为不足。..本申请课题关注实时和复杂化场景下的信息扩散研究,侧重其过程建模和抽取分析。首先,针对目前很多研究中信息扩散和用户行为等的割裂处理,本课题以概率图模型的方式设计综合的用户、信息和变迁扩散模型,以更好呈现扩散过程。其次,实时场景下的扩散关联抽取通过哈希和概述方法以达到快速演进关系抽取。最后,基于新的扩散结构模型和时序关联模式分析等工作,本课题设计更为准确的扩散预测方法。
中文关键词: 信息扩散;社交网络分析;图模型;预测算法
英文摘要: With the fast availability of social network and smart devices, we have witnessed a stunning volume of online information, which is also fast generated and spreaded quickly. A striking character is that, information like micro-blog, comment, interaction and location based posts can be generated in a real-time manner. At the same time, the associated features also make the processing more complex. Information diffusion focuses on temporal information and social network areas, modeling the content changes and users' behaviours. It attracts a lot of attentions, but the real-time and complex characters are usually ignored. ..This proposal focuses on the information diffusion research in real-time and complex scenarios, especially the modeling and analysis tasks. First, we propose a unified graph model to combine user, post and temporal features for better diffusion modeling, to overcome current isolated research. Second, we utilize the hashing and summarization techniques to cope with the real-time processing challenge of temporal diffusion mapping. At last, we design more accurate prediction algorithm based on the new diffusion model and temporal mapping work.
英文关键词: Information Diffusion;Social Network Analysis;Graph Model;Prediction Algorithm