| Summary: | 碩士 === 大葉大學 === 電機工程研究所 === 82 === Computing university course schedules is very hard. Course
scheduling is basically a multiple constraint satisfaction
problem, in which the determination of a solution is NP-
complete. The approaches oriented to operations research
simplified the problem to facilitate mathematical model
building and to reduce computation time. The AI/expert-system-
oriented approaches took advantage of powerful configuration
tools and supplied reasoning methods, but did not completely
solve the conflict problem between multiple constraints.
Via literature review and system analysis, this research
proposes simple heuristic rules to guide ''generate, test and
debug'' strategy to automate ng. A prototype system has been
developed, tested and evaluated. Course-scheduling by heuristic
rules can reduce computation time significantly. And the
huristic rules themselves are easier to understand than
mathematical models.
|
|---|
