Multiple case-based retrieval for university course timetabling problem

• Main Author: Hong, Siaw Theng
• Format: Thesis
• Language: English
• Published: 2016
• Online Access: http://psasir.upm.edu.my/id/eprint/69369/1/FSKTM%202016%2038%20IR.pdf; http://psasir.upm.edu.my/id/eprint/69369/
• Institution: Universiti Putra Malaysia

This thesis presents research for Case-based reasoning (CBR), a knowledge-based reasoning technique to solve university timetabling problem such as resource allocation for student’s course timetabling. CBR model’s was reviewed on Case-based Retrieval for timetabling discloses improvement that can be done to excel in accuracy and time consuming. From the review of past case-based retrieval techniques, a few concern is being investigate for the cases retrieval process such as the effectiveness of retrieval and time required to generate a comprehensive timetable. This research is aim to optimize the effectiveness of retrieval as well as generate a timetable in the shortest time possible with minimize violation. The case-based retrieval technique is further enhanced and improvised into a new algorithm known as Multiple Case-based Retrieval. The algorithm is combining separated distinct processes, with the combination of different functionalities: Prioritized Attributes, Frequency Grouping, and Value Difference Measurement. The algorithm was running on timetabling tests, comparing to classic Case-based retrieval and Genetic Algorithm for a wider comparison. Graphs are plot according to the readings from timetabling tests to show the result comparisons. Results from the experiments show the effectiveness and elapsed time to generate a timetable. Multiple Case-based Retrieval shows promising results in improving the effectiveness of case-based retrieval and also reduced the time required to generate a new timetable. This research summarize that the algorithm in retrieval is playing a very important role for an effective timetabling generator. Future research may concern to improve of the process of retaining cases, focus on case-based handling storage for generated cases for future review.