Reasearch And Implementation Of Team Formation Base On Heterogeneous Network

With the popularity of social networks,various collaborative programs and cooperation tasks are initiated on social networks,for example,the ZhuBaJie website,the Github website,and Freelancer websites.On the one hand,these websites are collaborative platforms on which professionals collaborate to complete projects;on the other hand,these networks form a social market for online,on which project sponsors post the tasks to find qualified professionals who meet the requirements to form a team.The purpose of the team formation algorithm is to get the team to meet the task requirements for the project sponsor based on the needs of the project tasks and personnel information,torecommended expert teams.With the rapid rise of collaborative networks,the number of talented people and the number of projects has been increasing.Therefore,the study of team formation algorithm is more urgent and more important.Traditional team formation algorithms only find the experts having skills required by the task to be a team,regardless of the efficiency of team members.In response to the above problems,this paper needs to study the formation of a team that meets the needs of skills and meet the constraints of low-cost communication,which is called the formation of a single-constraint team.First of all,this paper studies the communication theory model in the team,and still designs two kinds of quantification methods of communication cost according to the social network among experts:the maximum distance communication cost and the spanning tree communication cost.It is found that the team formation algorithm based on social network has the disadvantages of low time efficiency.This paper constructs a heterogeneous network that combines expert social networks and expert skills.Based on heterogeneous networks,this paper proposes the FindSkillDaim algorithm and the FindSkillAgg algorithm for different measures of communication cost.Both algorithms take the skills as their priority.The FindSkillDaim algorithm starts with a rarity to search iteratively in heterogeneous networks and select experts who lead to the least cost of the team when connected to other skills.FindSkillAgg algorithm draws on the idea of map conversion,the original map is converted to a complete map,and then reposition the skills needed to narrow the range before searching.As the number of candidates increases,candidates have their own strengths and skill levels.Teams also demand that specialists have a high degree of professionalism.Team formation problems have increased the level of skill constraints.In the process of multi-constraint team formation,after the team meets the skill requirements,it also needs to minimize the communication costs of the team and improve the skill level of the experts.In this paper,the classical algorithm is used to evaluate the level of expert skill,and then DPFW algorithm based on heterogeneous network is proposed.The random walk strategy is designed based on the level of expert skill,starting from each skill to walk,reaching the expert's skill of the probability of access,with the probability to select the team.In this paper,we achieve three sets of experiments on two datasets:DBLP,Github,to check the algorithms proposed on the correctness,constraint validity,and time efficiency.The experiment verify FindSkillDaim algorithm and FindSkillAgg algorithm has better binding and time efficiency,DPFW The algorithm can effectively solve the problem of multi-constraint team formation.