社会网络的团队形成问题已经逐渐成为社会网络分析以及数据挖掘领域的研究热点,现有团队形成问题的目标集中在查询一个成员间沟通代价最小的团队.在实际应用中,对于大规模任务通常需要按照模块进行任务划分,例如大型软件开发、大型科研项目等,因此完成任务的团队也需要进行分组.基于此需求,提出了社会网络上支持任务分组的团队形成问题,即从专家社会网络中查询出满足复杂任务分组且沟通代价最小的专家团队.该问题的查询输入不再是传统团队形成问题中的技能集合,而是输入一个分组任务图,证明了该问题是NP难问题.依据组织行为学中的团队沟通模型,定义了任务分组的团队沟通代价度量,并提出了基于不同贪心搜索策略的算法.采用真实数据集对所提出的算法进行了实验评估,实验结果表明依据不同的贪心策略实现的算法能够适用于不同的沟通代价度量方法,证明了算法的有效性.
Team formation problem in social network is gaining prominence in the research field of social network analysis and data mining.Previous study about team formation aimed at finding a team with the lowest communication cost.Some practical applications,such as large-scale software development and large-scale scientific research teams,usually need to divide a task.Based on this requirement,this paper presents a problem named grouping supported team formation in social network,which finds a team of experts to satisfy a complex grouping task and minimize the communication cost.The input of this problem is not a set of keywords of the traditional team formation problem,but a grouping task graph.Meanwhile,we also prove that this problem is NPhard.Based on team communication models in organizational behavior,we define communication cost criterions for measuring grouping task teams,and propose multiple corresponding greedy searching strategies.The experimental results on real datasets demonstrate that different search strategies are suitable for different communication cost criterions and prove the effectiveness of the proposed algorithm.