MISTA 2009 Programme
Plenary Speakers
Edmund Burke, University of Nottingham, UK
Title: Decision Support Methodologies for Healthcare Personnel Scheduling
Abstract: The challenge of building intelligent decision support methodologies to automatically generate solutions for healthcare personnel rostering problems has represented a series of demanding research questions, which have cut across Operational Research and Computer Science for over 40 years. Nurse rostering represents a key challenge in healthcare personnel scheduling. The field has generated significant international research interest in recent years. However, many of the papers that have appeared in the literature have addressed simplified versions of the problem. This presentation will discuss the goal of developing methodologies that can handle the complexity of modern real world scenarios. This goal represents a major scientific challenge. The presentation will discuss a number of advances and contributions within the context of closing the gap between academic research and real world practice in this area. I will also discuss a set of downloadable real world benchmark datasets and a suite of publicly available downloadable evaluation software that we have produced. This software and the benchmark suite of challenging and realistic problems is the first to ever appear in the nurse rostering research community. I will explore the use of decomposition techniques which have been integrated with variable neighbourhood search and which represent a particularly effective hybrid approach compared with commercially implemented methodologies. I will also present time pre-defined methods which are motivated by the goal of lowering the number of parameters. This methodology produces outstanding results on all the new benchmark problems that are mentioned above.
David Hine, Metropolitan Police, UK
Title: Football Sports Scheduling in the Real World (provisional title)
Abstract: A full abstract will be posted shortly but David will talk about his role in the Metropolitan in scheduling football fixtures (and other resources) in order to minimise costs as well as public disorder.
Moshe Dror, University of Arizona, USA
Title: ‘Strong’ - ‘Weak’ Precedence in Job Scheduling
Abstract: It is common knowledge that most applied combinatorial optimization problems with partial orders are NP-hard. Special cases of partial orders have been considered in great detail, permitting us to focus on an interesting precedence order distinction in scheduling. In this talk we will examine a partial order delineation of ‘strong’ and ‘weak’ precedence in chain and tree precedence structures.
We will present a summary of results regarding NP-hardness and polynomial time solvability for the distinction between ‘strong’ and ‘weak’ precedence in scheduling. In many cases, NP-hardness for weak precedence implies NP-hardness for the strong precedence. However, this is not universally true, and the ‘strong’ - ‘weak’ distinction is proper - at least for the case of chains. We primarily focus on the results for chains and trees but extend the 'strong’ - ‘weak’ precedence distinction to more general digraphs grounded in and motivated by actual real-life dispatching in multiprocessing systems that require stable schedules.
Raymond Kwan, University of Leeds, UK
Title: Case studies of successful train crew scheduling optimization
Abstract: The UK has a very large and complex passenger rail network divided into a number of franchises. Train crew scheduling is mission critical to the train operating companies, which would feel the pain if they have to schedule manually. Until recent years, none of the few attempts by these companies to adopt an automatic optimizing train crew scheduling system was successful. Typically, systems were commissioned based on their potentials but then they were unable to deliver schedules of acceptable quality, and they were eventually abandoned by the train companies. The TrainTRACS system has changed the scenario. ScotRail adopted TrainTRACS in 2003 and has remained an active user to date. This success has spread rapidly by the University of Leeds spin-out company Tracsis Plc, and now most of the major UK train operating companies are using TrainTRACS. This paper presents case studies of recent pilot trials for new Tracsis clients to adopt TrainTRACS, highlighting some special scheduling situations encountered.
Further information on the conference programme will be updated as it becomes available.
Regular Talks
-
Speaker name, organisation
presentation title
-
Speaker name, organisation
presentation title
-
Speaker name, organisation
presentation title
m