Optimal Meeting Points for Public Transit Users

Consider a group of colleagues going from their offices to their homes, via their preferred subway or bus routes, who wish to find k alternative restaurants to meet and which would minimize a given aggregate deviation distance from their typical routes. We call this the "k-Optimal Meeting Points for Public Transit" (k-OMPPT) query and present two approaches for returning provably correct answers for both SUM and MAX aggregate detour distances. Both approaches exploit geometric properties of the problem in order to refine the POI search space and hence reduce the query's processing time. Our experiments, using real datasets, compare the efficiency of both approaches and show which approach is preferable given the type of aggregate the group is interested in minimizing.