摘 要:該研究圍繞社交網絡的“群體行為形成與互動規律”這一科學問題,重點研究社交網絡群體行為與群體情感理論。在2013年度中,主要進行了如下4個方面研究:在社交網絡群體聚集機理及建模方法方面,針對群體聚集的誘因分析,從微博話題聚類、廣告用戶檢測、不良節點挖掘等方面進行了研究。針對群體聚集模型和凝聚力算法,從復雜社團分聚類算法、微博僵尸群檢測等方面進行了研究。針對群體行為屬性和共性需求機理,研究了群體隱私保護的共性需求和防護機制、策略描述及合成等問題;在社交網絡群體演化動力學行為方面,對群體行為演化模型及干預研究了微博信息傳播預警問題。在多維屬性的群體情感建模方法方面,從社交網絡用戶的個體人格結構、心理因素和經歷背景等要素與其行為特征、情感立場的關系,研究了個體的多屬性、多維度統計特性。在基于群體情感形成機理研究情感演化計算方法,開展了群體情感立場的量化、群體情感發現、基于異質網絡分析的個體行為信息的多屬性挖掘等方面的研究。基于以上研究要點,已發表研究論文28篇,其中SCI論文7篇,EI論文15篇。出版專著1部。申請發明專利4項,授權1項。
關鍵詞:群體行為 群體情感 重疊社區
Abstract:In this project, focusing on the scientific problem of group behavior formation and interactive principles, we study the group behavior and group emotion theory in the online social networks. In the year 2013, the project team developed the research mainly on four aspects. On the group aggregation principles and modeling methods aspect, aiming at incentive factors of group aggregation, we studied the topic aggregation in Weibo, detection of advertising users, and malicious user mining in SN. Aiming at group behavior attributes and common security requirements, we studied the group privacy protection policies and mechanisms. On the aspect of group behavior evolution, we studied the communication alert problem in the Weibo. On the multi-dimension attributes based emotion aspect, we studied multi-attributes and multi-dimension statistics characteristics of individual emotions from the relationship of individual personality, mentality, and background. On the group emotion formation principle base group emotion evolution calculation aspect, we studied the quantization of group emotion position, discovery of group emotion, multi-attributes mining of individual behavior information based on heterogeneous networks. Based on the research on the above four aspects, the research team published 28 papers, including 7 SCI indexed papers and 15 EI indexed paper, and 1 book. As to the patent aspect, we applied for 4 items and got 1 granted patent.
Key Words:Group behavior; Group emotion; Overlap community
閱讀全文鏈接(需實名注冊):http://www.nstrs.cn/xiangxiBG.aspx?id=49712flag=1