TeamFinder: A Co-clustering based Framework for Finding an Effective Team of Experts in Social Networks

A lot of research in social networks has been devoted in the recommendation of individuals. Most of the work has focused on finding appropriate people one in question, regarding the input social graph and their attributes. However, for many applications one is interested in finding a team of experts for a given query. For a given social graph and a task including a set of required skills, we study the problem of finding a team of experts that their expertises match the given task and the relationships among them represent how well team members work together. Our proposed framework, named Team Finder, first provide a grouping method to aggregate the set of experts that are strongly correlated based on their skills as well as the best connection among them. By considering the groups, search space is significantly reduced and moreover it causes to prevent from the growth of redundant communication costs and team cardinality while assigning the team members. Second, the identified groups are reconstructed as a bipartite graph an a Co-clustering method is applied to find the qualified clusters containing related experts and skills and finally, an adapted version of team formation method is used to discover the effective experts for accomplishing the given task. Team Finder method is developed in a scalable and effective manner and making it ideal for large scale applications. We conduct extensive experiments on DBLP co-authorship graph and the experimental results demonstrate the effectiveness and scalability of proposed method in practice and give useful and intuitive results.