Review - Mining Association Rules between Sets of Items in Large Databases.

Philip S. Yu: Review - Mining Association Rules between Sets of Items in Large Databases. ACM SIGMOD Digital Review 1: (1999) BibTeX

Review

This paper provides a mathematical formulation of the association rule mining problem, where the association rule problem tries to identify the set of items often appeared together in a transaction. It decomposes the problem into two subproblems. The first one is on the generation of large item sets based on support and the second one is on deriving the association rules from the large item sets based on confidence. This pioneer work has provided an elegant problem formulation that transforms an abstract problem into an algorithmic problem. It opens up a new area for future research. In additional to potential research opportunities for faster mining algorithms and model extensions, I was most intrigued by the issues on the statistical significance of the rules by considering alternative measures other than support and confidence such as collective strength which is a correlation type measure, and also on the on-line interactive generation of the rules to give users more control on what rules to generate and how to specify the parameters.

References

[1]: Rakesh Agrawal, Tomasz Imielinski, Arun N. Swami: Mining Association Rules between Sets of Items in Large Databases. SIGMOD Conference 1993: 207-216 BibTeX

BibTeX

Digital Review - DBLP: [Home | Search: Author, Title | Conferences | Journals]