Abstract: The present team formation researches in social network mostly take 0-1 rule to measure expert skills.Aiming at the situation that people often utilize the natural language to describe expert skills,this paper proposed a social team formation method based on fuzzy multi-objective evolution.This method focuses on how to find out the appropriate individuals from the expert social network to form a team with certain size and achieves the optimization between communication cost and team performance under the uncertainty circumstances.In this method,the precise parameters represented by 0-1 rule are replaced by fuzzy language variables to describe expert skill.The concept of team performance is used to measure team capability.Because the standard SPEA2 algorithm has slow convergence at the initialevolutio-nary stage,this paper introduced AEL strategy to generate individuals with good characteristics.Considering the ambi-guity of expert skills,this paper also proposed a fine-grained Dominance judgment as the new rule of judging the dominance relationship of individuals.The simulation results show that the improved algorithm converges fast and obtains good quality approximate PF,which can be successfully applied to solve the team formation problem.

Key words: Evolutionary algorithm, Fuzzy language variables, Social network, Team formation