Methodological assistance for integrating data quality evaluations into case-based reasoning systems

Abstract

Case-based reasoning systems are used in more and more problem-solving domains supporting the long-term reusing and storing of experience. The performance of these systems essentially depends on the quality of the experience items in their knowledge base, represented as data. Defects in the quality of these data may interfere with the system's performance. By means of inspection and review the data quality is measured, evaluated, assured and improved. To support these activities in a case-based reasoning system, data quality criteria and control processes are required. Previous work in the field of data quality in case-based reasoning remains at a comparatively coarse-grained level. Existing approaches mostly do not provide sufficient methodological assistance in defining fine-grained quality criteria or designing and implementing control processes for the measurement and evaluation of the data quality. Therefore this paper proposes two approaches for methodological assistance in developing data quality inspections and data quality management for case-based reasoning systems. © Springer-Verlag Berlin Heidelberg 2007.