Preliminary Evaluation of Discovered-Rule-Filtering Methods

Data mining systems semi-automatically discover knowledge by mining a large volume of data, but discovered knowledge is not always novel to users. We discuss a discovered rule filtering method to filter rules discovered by a data mining system into ones that are novel to the user by using information retrieval results from the Internet. We have two methods; the micro view and the macro view methods, to achieve discovered rule filtering. In the micro view method, we extract keywords from a discovered rule and rank the rule referring to the number of hits when the keywords are submitted to the MEDLINE database. In the macro view method, we first retrieve documents by submitting every pair of the extracted keywords and then form keyword clusters according to the results. We evaluated the methods by sending out a questionnaire to medical students. The evaluation indicates that the macro view method is promising as a discovered rule filtering method.