Realistic team formation using navigation and homophily

This paper proposes an algorithm for selecting a team of experts from a social network, which is represented by a weighted undirected graph. Each node in the graph represents an individual who has one or more skills. We introduce the concept of realistic team formation and its associated constraints and assumptions. The main constraint is a local view of the network resulting in an absence of pre-computed network statistics. Nodes in the network are assumed to connect to people with whom they share common skills. Thus, this paper proposes a navigation based team formation algorithm that makes use of homophily to recruit candidates for the team. The objective is to create an effective team that can carry out a specific task. Since none of the related work has had a local view of the network, we use BfsRecruiter, a BFS based navigation algorithm as our baseline comparison. We implemented the proposed algorithm and ran many simulation experiments to measure the communication cost and effectiveness of the proposed algorithm.