The popularity of association rules has resulted in several variations being proposed. In each case, additional attributes in the data are considered so as to produce more informative rules. In the context of active mining, different types of rules may be required over a period of time due to knowledge needs or the availability of new attributes. The present approach is the ad-hoc development of algorithms for each variant of rules. This is time consuming and costly, and is a stumping block to the vision of active mining. We argue that knowledge needs and the changing characteristics of the data requires the ability to re-specify the type of rules to rediscover over time. This paper proposes a novel approach to specify the "how-to" of mining different rule variants without the cost of developing new algorithms. Called the VDL, it is SQL-like and has the expressive power demonstrated by our examples, some of which are classical and others novel. We also give a discussion on the theoretical model underpinning our proposal.
Databases and Information Systems | Numerical Analysis and Scientific Computing
Data Management and Analytics
International Workshop on Active Mining AM 2002, in conjunction with the IEEE International Conference on Data Mining ICDM 2002, 9-12 December
City or Country
Maebashi City, Japan
ONG, Kok-Leong; NG, Wee-Keong; and LIM, Ee Peng.
VDL: A Language for Active Mining Variants of Association Rules. (2002). International Workshop on Active Mining AM 2002, in conjunction with the IEEE International Conference on Data Mining ICDM 2002, 9-12 December. 1-6. Research Collection School Of Information Systems.
Available at: http://ink.library.smu.edu.sg/sis_research/905
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.