Content-Based Conference Scheduling Optimization

A scientific conference is an activity organized by a community of researchers aimed at offering a platform for knowledge sharing and scientific exchange. The planning of a conference includes the creation of the program schedule where decisions about track composition, in terms of talks, and the temporal allocation of those tracks need to be considered. Designing such a schedule represents a difficult task in which aspects as content similarity and the overlapping of similar tracks should be taken into account. In this paper a two-phase solution approach is proposed for helping planners designing schedules. In first place, talks are grouped into tracks upon the basis of talks content. Then, a timetable for the tracks is created by group them over the basis of content similarity, avoiding those with similar content to be scheduled in parallel. Both problems demonstrate to require a considerably computational effort; consequently, advanced algorithmic approaches as column generation and GRASP are exploited with the purpose of finding near-optimal solutions in a short computational time.