Managing the tabu list length using a fuzzy inference system: An application to exams timetabling

In this paper we present an application of Tabu Search (TS) to the exams timetabling problem. One of the drawback of this meta-heuristic is related with some parameter that can affect the performance of the algorithm and must be tuned. One of such is tabu tenure which is much problem dependent. Our goal is to have a automatic process of managing the memory which is important for practical applications. We can in general consider that the degree of expertise of the users (university staff for instance) is low. To automatically manage the tabu tenure we employ techniques used in fuzzy decision expert systems. We developed a fuzzy inference rule based system (FIRBS) to handle the tabu tenure based on two concepts”frequency” and”inactivity”. These concepts are related to the number of times a move was attempt and the last time it was called. We have implemented the simplest form of a tabu search in order to evaluate this new feature. Computational results show that using a FIRBS improves the performance of Tabu Search.