Abstract
Data mining is used to discover hidden patterns or structures in large databases. Association rule induction extracts frequently occurring patterns in the form of association rules. However, this technique has a drawback as it typically generates a large number of association rules. Several methods have been proposed to prune the set of extracted rules in order to present only those which are of interest to the domain experts. Some of these methods involve subjective analysis based on prior domain knowledge, while others can be considered to involve objective, data-driven analysis based on numerical measures that provide a partial description of the interestingness of the extracted association rules. Recently it has been proposed that ontologies could be used to guide the data mining process. In this paper, we propose a hybrid pruning method that involve the use of objective analysis and subjective analysis, with the latter involving the use of an ontology. We demonstrate the applicability of this hybrid method using a medical database.
Original language | American English |
---|---|
Journal | Default journal |
State | Published - Jan 1 2011 |
Keywords
- Association rule induction
- Data mining process
- Healthcare
- Ontologies
- Objective measures
- Subjective measures
- Interestingness
Disciplines
- Business