Research on Team Formation Method in Social Network

Xiao-Qin XIE,Shuai HAN,Min CHEN,Zhi-Qiang ZHANG,Yi-Jia LI,Hai-Wei PAN
DOI: https://doi.org/10.11897/SP.J.1016.2017.00712
2017-01-01
Abstract:How to use social network information to find a team with efficient collaborating performance and good quality has become a hot research issue in social networks.Current researches usually use 0-1 value to measure whether an individual has a skill or not while in practical applications the way to judge whether an individual has certain skills will largely influence the efficiency of the team to complete the task.Moreover, most present studies only focus on considering the number of the past collaborated tasks and do not deeply explore different social relationships between individual as well as the individual's own other properties while calculating the strength of relationships between individuals in social network.These factors immensely influence social relationships and team cooperation between individuals.To solve the above problems, firstly, we give the problem definition and the related concepts as well as the definition of skill contribution and make use of a relation model to calculate the relationship between individuals which takes into consideration the categories of social connections and the properties of individuals.Secondly, according to the different forms of team, we study two types of team formation problems: team formation problem without a leader and team formation problem with a leader.Then we propose Minimum Covering Steiner-based Team Forming Algorithm and Minimum Set Covering-based Team Forming Algorithm to search for the best team without a leader separately and propose Minimum Leader Distance based Team Forming Algorithm to search the best team with a leader.Finally, we use DBLP as dataset and conduct experiments to verify the feasibility and effectiveness of all of the above methods, and compare our methods with existing algorithms from the following aspects: team cooperation cost, the cardinality of the team, team connectivity and the comparison results of social influence factors to algorithms.The experimental results prove the feasibility and effectiveness of our methods.
What problem does this paper attempt to address?