Modelling and Solving the Italian Examination Timetabling Problem using Tabu Search

The Examination Timetabling problem (ETTP) regards the scheduling for the exams of a set of university courses, avoiding the overlap of exams of courses having common students, and spreading the exams for the students as much as possible. Carter et al [3] provide a set of formulations for ETTP, which differ from each other on some components of the objective function, along with a set of benchmark instances [2]. Formulations and benchmarks by Carter have stimulated a large body of research, so that many researchers (see, e.g., [1,5,4]) have adopted one of the formulations of Carter (or a variant of them), tested their algorithms on the benchmarks, and also added new instances. At present, all instances and the corresponding best results are published on the Web [8]. Unfortunately though, the real-world formulation that applies to our institution and, up to our knowledge, to most Italian universities, differs substantially from Carter’s and similar ones. Therefore, in order to solve our practical problem, we have been forced to model and solve the specific situation of our university almost from scratch. In this paper, we present an ongoing research on the development of a solution software for the ETTP at the school of Engineering of the University of Udine. In details, we present the problem modelling, the structure of the real instances, and the solution algorithm, which is based on Tabu Search [7] and on the interleaving of different neighbourhood relations, in the spirit of [6].