Sunday, May 25, 2008

[Reading] Mining Association Rules between Sets of Items in Large Databases

This paper presents a system to find the association rules from a large number of transaction records. At the time of publication, the problem and the solution of this field might be very twilit and this should be a very pioneering work.

After properly formulating the problem, the problem can be decomposed into 2 sub-problems, and the trivial solutions become obvious. However, the straightforward algorithm is too expensive, and thus the authors provide many acceleration methods. I personally like the analysis at the memory analysis part.

The main difficulty in reading this paper is that the writing style is very old (or at least very different to the papers in my field). Also the results are obtained from a dataset without ground truth.

No comments: