Top-k team synergy problem: Capturing team synergy based on C3

Why is a team greater than the sum of its members’ capabilities? Forging a team depends upon solid collaborations among the team members amalgamated with each member's abilities. These two aspects bring a challenge in finding the right mix of members with a novel notion of Synergy (Sy) from graph G. This paper has three main goals: (i) introducing the notion of Team Synergy Problem (TSP) and proposing a novel Sy function, (ii) identifying the intrinsic structure of G for predicting potential Sys, and (iii) developing a top-k Team Synergy Algorithm (TSA). Specifically, we formulate the TSP by embedding three essential elements (C3); Communication, Cooperativeness, and Complementarity, into the Sy function to quantify the Synergy between adjacent experts and construct a Synergy graph, G_S. We prove that the TSP is NP-hard and propose TSA to form top-k teams from G_S within a budget B. TSA uses PSEUDO-STAR configurations to prune instances efficiently. Moreover, it uses a tensor decomposition method, RESCAL, to exploit the tensored Synergy graph, G_S, to predict the potential Synergies on the unknown edges and recommend new teammates to a given team. The experimental results on four real datasets have shown that TSA significantly outperforms the state-of-the-art algorithms.