The use of collaboration distance in scheduling conference talks

Abstract

Several bibliographic databases offer a free tool that enables one to determine the collaboration distance or co-authorship distance between researchers. This paper addresses a real-life application of the collaboration distance. It concerns somewhat unusual clustering; namely clustering in which the average distances in each cluster need to be maximised. We briefly consider a pair of clusterings in which two cluster partitions are uniform and orthogonal in the sense that in each partition all clusters are of the same size and that no pair of elements belongs to the same cluster in both partitions. We consider different objective functions when calculating the score of the pair of orthogonal partitions. In this paper the Wiener index (a graph invariant, known in chemical graph theory) is used. The main application of our work is an algorithm for scheduling a series of parallel talks at a major conference.