University of Otago logo. Computer and Information Science Seminars

Seminar Homepage


Nathan Rountree, Department of Computer Science


Generating "Rare" Association Rules


Archway 2 - 1:00 pm, Friday 18 September


Association Rule Mining is an attempt to answer the question, "What things typically appear together in my database?"

The most famous method of generating answers to this question is the "Apriori" algorithm, which produces statements like "70% of people who buy milk also buy eggs."

And that's about as exciting as Apriori gets, because it only looks at things that occur frequently (like eggs) and not infrequently (like emeralds). To find rules about infrequent things, one has to set Apriori's "support" parameter so low as to get an explosion of rules, possibly larger than the original database.

My former doctoral student Yun Sing Koh and I have just published a book-length collection of new research papers on generating association rules for rare items. In this talk, I'll outline our particular approach to the problem, and give a quick overview of things that our other authors wrote about.

Last modified: Thursday, 17-Sep-2009 10:56:21 NZST

This page is maintained by the seminar list administrator.