A Realistic Criterion for Team Formation in Social Network

Team Formation Problems in Social Networks (TFP-SN) has become one of the most popular areas in Social Network Analysis (SNA). Researchers usually use a standard framework to solve these problems. A network of the experts is modeled by graphs; the nodes of which are representative of the experts and the edges between them represent the communications between them. After that, based on the costs of each expert and also the level of his relationship with other team members, the final team could be formed with the lowest cost. Therefore, they generally face with two objective functions, and both of which must be optimized in such problems. In previous studies, researchers have tried to provide a variety of objective functions for personal, communication, or both costs that lead them to a more efficient team at a lower cost. However, the current objectives are not able to take many other human considerations into account, resulting in sub-optimal teams. In this paper, we show how considering and formulating one of such human considerations can form teams with lower costs. More precisely, we first introduce a new objective function to calculate personal costs and then formulate one of the human considerations, which eventually results in removing experts with unusual salaries in the final team. The experimental results show that applying the new objective function, as well as the new consideration of human selection, can lead to superior results in reducing personal costs, communication costs and, therefore, the total cost of a team.