Team building by data clustering with constraints

Making cooperative teams from multiple participants is a common exercise during the beginning of teamwork collaboration. Due to the priori knowledge among participants, several team building constraints can be formed. For example, it might be required to place participants from the same organization in the same group. In this research, we consider the cooperative team building as a constrained clustering problem. Clustering algorithms equipped with instance level must-link (ML) and cannon-link (CL) constraints are known to improve the efficiency and accuracy on handling data clustering. This paper presents the result of applying the constrained clustering algorithm to accommodate the complete must-link (CML) constraints, a special case of ML constraints, in which all participants should be pre-grouped with other members. The result shows the CML clustering can be used to group teams by pre-group requirements with promising computational performance.