R. Kuusik, I. Liiv and G. Lind, An Efficient Method for Post Analysis of Patterns, Artificial Intellegence and Applications, 2005, 453, pp.294-309, (pdf)
Tallinn University of Technology, Department of Informatics, Raja 15, 12618 Tallinn, Estonia
Abstract
Finding and extracting frequent patterns is one of the most important tasks
in data mining, therefore various algorithms have been introduced over time.
Unfortunately, when the sizes of datasets increase, completely different and new problems
arise. Even if we are able to extract the IF
THEN rules in a reasonable time, it is
possible that the algorithms will find millions of patterns. Interpretation of all of them
would be a grievous baffling problem for even a team of analysts. In this paper we
describe a method we have used for post-analysis of patterns. The basic idea is presented
with an example and the rules for result transformation are given, making it possible to
apply standard querying tools. Although we have implemented it as an extension to
generator of hypotheses, it would also give reasonable results with other rule extracting
methods.
Keywords
Pattern post analysis, data mining, generator of hypotheses, and monotone system theory