Hybridisations within a Graph Based Hyper-heuristic Framework for University Timetabling Problems

A significant body of literature has explored various research directions in hyper-heuristics (which can be thought as heuristics to choose heuristics). In this paper we define a graph based hyper-heuristic (GHH) framework and compare the performance of a number of local search based searching algorithms as the high level heuristics. They are used to search upon permutations of low level graph heuristics rather than solutions within the GHH framework. We also investigate efficient hybridisations between the GHH and local search methods. Aiming at gaining an in-depth understanding on this new technique, we address and discuss the fundamental issues concerning the neighborhood structures, the characteristics of the two search spaces (namely the search spaces of the graph heuristics and the actual solutions). Issues of exploring by the high level search and exploiting by the local search within the hybrid GHH are also addressed. These, to our knowledge, are entirely novel directions in hyper-heuristics. The comparisons among different high level heuristics within the unified framework suggest that the role high level search play is to explore as wide search space as possible within a limited time using simple and iterative methods. The efficient hybridisations within the GHH obtained competitive results compared with the best published results for both benchmark course and exam timetabling problems, demonstrating its efficiency and generality. Possible extensions upon this simple yet general graph based hyper-heuristic are also provided.