Team Formation

Overview

Team Formation tackles the challenge of assembling optimal teams of experts from large collaboration graphs. From 2019-2022 we explored this problem as a complex instance of keyword search in graphs, requiring multiple criteria—skill coverage, productivity, communication efficiency—to identify the best subgraph. Our research produced novel graph-neural-network architectures that surpass prior state-of-the-art approaches.

Key Contributions

• Formulated multi-objective team-formation as a graph-search learning task.
• Designed GNN models integrating skill semantics, collaboration history, and communication cost.
• Demonstrated significant performance gains on real-world co-authorship and enterprise datasets.

Collaborators

• AT&T Chief Data Office
• York University

Output

• Multiple publications: (missing reference), (missing reference), (missing reference), (missing reference), (missing reference), (missing reference), (missing reference)

Our GNN-driven framework enables organizations to quickly identify high-impact, well-balanced teams for complex projects.