Research On Team Formation Based On Social Collaborative Computing

Teamwork,as the primary working form,is playing an increasingly significant role in modern organizational management.Good teamwork makes it easier to obtain glorious achievement compared to single-handedness.With the multidisciplinary crossing and penetration and the growing complexity of research,the phenomenon of interdisciplinary cooperation among people with different background knowledge is increasing.Internet social platforms have been widely used at the same time,online collaborative communication has become a trend in the future.The team formation under the social network became a research hotspot.In response to this problem,the innovations of this thesis mainly include the following aspects.(1)In the existing research,the measurement of communication cost for any two non-directly connected nodes in a social network is mostly based on the shortest path,which may result in two nodes that are not directly connected may have a lower communication cost than directly connected nodes,and it is not profound enough because of losing sight of the individual's attribute.Aiming at this problem,this thesis proposes a new model to measure the strength of individual relationship.The strength of social relationship is described from the aspects of individuals interaction familiarity and attributes similarity.The communication cost between individuals is defined as the reciprocal of relationship strength,and team formation with communication cost optimization is studied.(2)The existing research designed adaptive algorithms for different ways of defining the communication cost of team.To make it more universal,this thesis designs the discrete form of imperialist competitive algorithm and introduces the genetic operator to replace the original assimilation mechanism.In addition,in order to strengthen the interaction between the empires,a crossover operator is applied to the imperialists to seek a better solution,proximity detection and mutation operator are able to increase the probability of escaping from the local optimal solution.(3)At present,0-1 rules and skill grading are adopted by most research to measure individuals' skills.However,it is rare to require a skill to reach a certain exact value in reality.Therefore,the problem of team formation in social networks under a fuzzy environment is proposed.The individuals' skills level are described with qualitative linguistic variables such as “good” and “poor”,and then convert them into triangular fuzzy numbers.Given a task that require multiple skills,select a number of members from the social network to form a team to achieve communication cost and team performance optimization.On the basis of the standard SPEA2 algorithm,combined with the ambiguity of members' skills,the thesis proposes a fine-grained dominance judgment,and introduces the archive elite learning mechanism to generate internal population individuals.(4)In order to prove the feasibility and effectiveness of the proposed algorithm,the DBLP and information on google scholar web pages are used as experimental data.First,the validity of the relationship strength model is verified.The RarestFirst algorithm,the BEST-SD algorithm and the RarestFirstGreedy algorithm are compared with the ICA-TF algorithm,four kinds of communication cost are selected as the criteria for evaluation.The experimental results show the efficiency of the ICATF algorithm.In addition,the improved SPEA2-AL algorithm has a fast convergence speed and good distribution diversity,which can effectively solve the team formation problem.