Discovered Rule Filtering Using Information Retrieval Technique

A data mining system can semi-automatically discover knowledge by mining a large volume of data, but the discovered knowledge is not always novel and interesting to the user. We propose a discovered rule filtering method to filter rules discovered by a data mining system and to produce ones that are novel and interesting to the user by using information retrieval technique. In the method, we rank discovered rules according to the result of information retrieval from the Internet. In this paper, we show the steps of discovered rule filtering by using a concrete example of clinical data mining and MEDLINE document retrieval. Preliminary results show that this method has merits in not only filtering discovered rules but also providing a new viewpoint to the rules to give a chance to invoke a new data mining process.