University of Pittsburgh


Director of Decision Support and Information Technology
Friday, October 2, 2015 - 1:00pm - 1:30pm

Class scheduling in higher education, known in the literature as “timetabling in higher education”, is a complex business process that involves many people across an institution during several weeks or months every year. Timetabling is also considered a difficult problem from the theoretical perspective. There is a wealth of research on the topic focused mostly on development of optimization algorithms with early works starting in the 1950s and a noticeable increase in the volume of published works since the early 2000s. The proposed architecture and methodology are intended to address core reported gaps that exist between research on the topic and the needs of higher education institutions. Actual enrollment data from the University of Pittsburgh main campus is used as case study.


The core components of the proposed architecture are specified by: First, modelling the problem as an association rules analysis where the sets of courses that individual students are enrolled in during an academic term are treated as transactions. This analysis renders an initial set of combinations of courses of interest called course itemsets. A backtracking algorithm called MASAI-1 is proposed to determine the maximum number of seats available per itemset. MASAI-1 is a novel approach to the identification of itemsets of interest as it uses information that is not available in the transactional data. Second, in order to facilitate deeper analyses that consider the relationships between course itemsets, the problem is modelled as a multi-mode graph that includes information obtained with the association rules analysis and MASAI-1. A Generalized Clique Percolation Method (GCPM) is proposed to enable the identification of overlapping and hierarchical communities in graphs/networks. Third, the elements that would form the core of a socially translucent environment that is based on the two previous components are discussed. 

Copyright 2009–2020 | Send feedback about this site