基于数字足迹的社群智能关键技术研究

项目名称： 基于数字足迹的社群智能关键技术研究

项目编号： No.61472283

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 计算机科学学科

项目作者： 张大强

作者单位： 同济大学

项目金额： 85万元

中文摘要： 随着移动计算、社会计算和人类行为研究的逐渐融合，社群智能这一研究方向正日益兴起。然而，数字足迹的海量性和异构性以及人类行为的复杂性给社群智能的研究带来严峻的挑战。本课题以社群智能为研究对象，以建立一套社群智能的基本理论和方法为研究目标，以社群的识别、演化和应用为主线，以先个体后群体为辅线展开研究。主要工作有：1）基于Bayes理论和滤波跟踪等方法的数字足迹智能抽取和融合技术，包括身份识别和数据融合；2）提出异质和区域相遇间隔时间，分析个体移动模式，建立高阶Markov链模型；3）扩展完全子图渗流法来动态识别社群，消除不识别单个社群和社群结构等缺点；结合个体移动模式，利用上述概念，分析群体移动的时空特征，建立群体移动模型；4）研究基于社群的报文传输机制，进而研究社群智能服务的应用模式。本项目的研究成功，将建立一系列社群智能的关键技术，取得原创性的成果，力争为推动社群智能的发展做出重要贡献。

中文关键词： 数字足迹；社群智能；移动社交网络；数据分析；情感计算

英文摘要： With the proliferation and fusion of mobile computing, social computing, and human behavior study, Social Community Intelligence (SCI) is emerging. However, it presents significant challenges in SCI owing to the heterogeneity and large-scale volumns of digital footprints, and the complexity of human behaviors. By conducting intensive and extensive research on SCI, this proposal aims to establish a series of fundamental theories, principles, and application methodologies for SCI. It follows the lifespan of SCI to organize the study, involving SCI discovering, evolution and application. It also complies with the logic of individuals-first-and-community-second. This proposal firstly analyzes the large-scale digital footprints that reveal the information of individual behaviors, and builds a mobility model using the high-order Markov chain. Then, it goes to the study on the collective movement. The main contributions of this proposal are four-fold. 1) It will present a couple of mechanisms using Bayes theory, filtration tracking and other techniques to intelligently extract information from digital footprints and perform data fusion. This step consists of two substeps: object identitifcation and data fusion. 2) It introduces two concepts - - heterogeneous inter-contact time (HICT) and regional inter-contact time (RICT). By analyzing the temporal and spatial features of indivuals' movement using HICT and RICT, this proposal proposes a mobility model using the high-order Markov model. To explore the RICT, it introduces the concept of region, and takes advantage of spatial analysis mechanisms (e.g., Moran's I and Geary's statistics) to idenfiy regions. 3) It will extend the clique percolation method (CPM) to discover the overlapping social community structure. The proposed method attempts to bring new abilities to CPM, i.e., discovering communities that are composed of only one member, and recognizing the community layers. 4) It will come up with a mechanism for message delivery. By this study, it also plans to show a manner that SCI services and applications server users. In summary, the sucessful execution of this proposal will generate some enabling mechanims of social community intelligence.By our original and creative efforts, we expect to promote the development of social community intelligence.

英文关键词： Digital Footmarks;Social Community Intelligence;Mobile Social Networks;Data Analysis;Context Awareness