Comparison between objective interestingness measures and real human interest in medical data mining

Authors:
Miho Ohsaki;Yoshinori Sato;Shinya Kitaguchi;Hideto Yokoi;Takahira Yamaguchi
Affiliations:
Shizuoka University, Faculty of Information, Shizuoka, Japan;Shizuoka University, Faculty of Information, Shizuoka, Japan;Shizuoka University, Faculty of Information, Shizuoka, Japan;Chiba University Hospital, Medical Informatics, Chiba, Japan;Shizuoka University, Faculty of Information, Shizuoka, Japan
Venue:
IEA/AIE'2004 Proceedings of the 17th international conference on Innovations in applied artificial intelligence
Year:
2004

Citing 0
Cited 3

Modeling interestingness of streaming association rules as a benefit-maximizing classification problem

Knowledge-Based Systems
An information value based approach to design procedure capture

Advanced Engineering Informatics
Evaluating learning algorithms with meta-learning schemes for a rule evaluation support method based on objective indices

PKAW'06 Proceedings of the 9th Pacific Rim Knowledge Acquisition international conference on Advances in Knowledge Acquisition and Management

Quantified Score

Hi-index	0.00

Visualization

Abstract

This research empirically investigates the performance of conventional rule interestingness measures and discusses their availability to supporting KDD through system-human interaction in medical domain. We compared the evaluation results by a medical expert and that by selected measures for the rules discovered from a dataset on hepatitis. Recall and ?2 Measure 1 demonstrated the highest performance, and all measures showed different trends under our experimental conditions. These results indicated that some measures can predict really interesting rules at a certain level and that their combinational use in system-human interaction will be useful.