Social Network Mining and Search
Irwin King (The Chinese University of Hong Kong, Hong Kong)
Christos Faloutsos (CMU, Pittsburgh, PA, USA)
Chin-Yew Lin (Microsoft Research Asia, China)
Cong Yu (Yahoo! Research, Sunnyvale, CA, USA)
Philip Yu (University of Illinois at Chicago, USA)
Social networking services and sites are florishing around the world, and they are changing the landscape of how we communicate and interact with each other. Sites such as MySpace, Facebook, and Orkut attract more than ten-billion visits per day. According to the statistics of Google OpenSocial, about one billion users have registered on social networking sites. With the enormous amount of social interaction data, social network analysis has emerged as an important technique to mine relationships among users, and the mined information can be used on a range of applications from improving information retrieval quality, conducting viral marketing, to finding domain-specific experts. This panel will discuss two data engineering research issues: social-network data mining and social signals for personalizing search. A social network can be formulated into a large graph of nodes, where each node represent a user and a link represents relationship between users. Though some traditional graph mining techniques can be employed to mine user relationship, they are not scalable to deal with millions and billions of nodes. Besides, a relationship can be beyond the traditional “relevance” model. For instance, a relationship can reflect influence and degree of influence; and relationship can change according to the “state” of nodes. This panel will discuss what social signals are potentially useful and how the signals can be mined in a scalable fashion. In particular, the panel will discuss how social signals can help improve Web search.