Download presentation
Presentation is loading. Please wait.
Published byOpal Hensley Modified over 9 years ago
1
University Course Timetabling with Soft Constraints Hana Rudova, Keith Murray Presented by: Marlien Edward
2
Introducing the Authors Hana Rudova : Faculty of Informatics, Masaryk University Czech Republic Keith Murray: Space Management and Academic Scheduling, Purdue University, USA
3
Introduction Using CLP to construct a huge automated timetabling system at Purdue university Declarative nature of problem description Constraint propagation technique for reducing search space Features of the problem Demand driven Different variables have different preferences Over constrained
4
Problem Description Class combination meeting student’s needs Classroom allocation respecting both instructional requirements and faculty preferences (e.g. time requirements or class preference) Students sectioning to be easy to measure the desirability and undesirability of classes overlap
5
Managing Soft Constraints Preference constraint solver Handles preference values for preference variables Zero preference: complete satisfaction Higher preferences: expresses the degree of violation Values not present in the preference variable domain: infinite preference **The preference solver maintains a cost variable for each preference variable having the best preference variable as its lower bound
6
Managing Soft Constraints (contd) Types of Soft Constraints: 1.Unary on time variables faculty time preference 2.Unary on classroom variables faculty preferences on the classroom selection 3.Binary for joint enrollment between two classes
7
Searching Aim: is to search for a complete assignment of preference variables to achieve the best satisfaction of all constraints Technique: initial partial assignment of the variables then repair it such that most of the variables are assigned values
8
Managing Hard Constraints Some useful constraints: disjoint2 (+Rectangles) To ensure that each class is assigned to just one suitable classroom cumulative (+Starts, +Durations, +Resources, ?Limit) To ensure that a resource can run several tasks in parallel
9
Conclusion Solution included: New solver for soft constraints Repair search for finding a solution Some constraints for stronger constraint propagation Future work: Extension of problem solution Improving the preference solver New approach for initial student sectioning
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.