Mykola Galushka and David Patterson, University of Ulster
Temporal features are an important characteristic of real world data, yet often they are ignored within the context of case-base reasoning (CBR) systems. This is largely due to the difficulties of handling and maintaining temporal knowledge effectively. We present a novel, domain independent approach, called D-HST, designed for case retrieval in CBR, we assess its limitations and evaluate modifications designed to address these shortcomings. We show the approaches to be both efficient and competent when indexing real world case-bases, compared to the commonly used R-tree indexing technique.