|
Abstracts
Contributed Talks
Implementation of a University Final Examination Timetabling System
Mehmet Sevkli, Arife Burcu Colak, and Mazhar Unsal
The university final examination timetabling problem is investigated.
Most previous investigators have resorted to heuristic methods of
solving the university final examination timetabling problem because
of the difficult combinatorial nature of this problem considering the
voluminous amount of decision variables and constraints involved. A
nonlinear integer program formulation of the problem is presented in
the present study which is then transformed into a linear integer
program. The resulting problem is solved using a commercial software
package. Results from the present study together with the limitations
of the present method are discussed by means of a practical example.
Determining Feature Weights in a Case-Based Reasoning Approach to Nurse Rostering
Gareth Beddoe and Sanja Petrovic
We present a method for capturing nurse rostering experience and
adapting it to solve new problems using the Case-Based Reasoning
paradigm. This method allows rostering knowledge to be stored
implicitly rather than by using inflexible sets of IF-THEN rules.
Cases are retrieved from the case-base using characteristic feature
values. The problem of automatically assigning weights to these
features is investigated. These weights must represent the
relevance and relative importance of the features based on a set of
supplied training cases. We discuss various approaches and present
an automated feature weighting algorithm tailored to the case-based
rostering method.
Building Quality Timetables with Integer Programming
Sophia Daskalaki and Theodore Birbas
We propose an IP formulation for certain types of timetabling problems.
Students attend general education, specialized and elective courses.
Forming timetables for schools with such system is difficult, because
students form groups to attend some courses and then split to form
different groups. When compactness in student timetables is required,
the problem becomes even harder. With our model we introduce a number of
0-1 variables, very primitive in structure, thus flexible in modeling.
The model obeys all hard functional constraints imposed by the
educational system. In addition, quality issues are addressed to create
more satisfying timetables for the users.
Scheduling the Italian Football League: an ILP-Based Approach
Federico Della Croce and Dario Oliveri
We consider the problem of scheduling the Italian Football League. This
problem consists in finding a round robin schedule taking into account
home-away conditions, twin-schedules for teams belonging to the same
home-town, specific balanced calendars constraints due to cable
televisions requirements and so on. We adapt the ILP-based approach
proposed by Nemhauser and Trick to schedule a College Basketball
Conference that considers however only half of the teams involved here.
The proposed approach allows generating within few minutes several
different reasonable ideally balanced calendars minimizing the total
number of home-away breaks and satisfying various other operational
constraints.
|
