A Quality Model For The Evaluation Of Decision Support Systems Based On A Knowledge Discovery From Data Process

Providing final users with confident knowledge to help them make the right decisions is the goal of using a decision support system based on a knowledge discovery from data process (DSS/KDD). Some failures can be found in such systems especially when the mined knowledge is unconfident, or if the system is hardly usable. The objective of this study is to define a quality model (QM) which ensures a global evaluation of DSS/KDD that generates association rules. The proposed QM evaluates the DSS/KDD regarding three dimensions: utility, usability and interestingness. It defines a set of criteria and allows the measurement of a DSS/KDD quality evaluation. To validate the proposed approach, a prototype has been developed. Weka and a DSS/KDD in the healthcare domain were assessed drawing on 20 users who participated in the evaluation process. Results have shown that a user-centred QM leads to a better quality of such systems.