On-Demand and Model-Driven Case Building Based on Distributed Data Sources.

The successful application of Case-Based Reasoning (CBR) depends on the availability of data. In most manufacturing companies these data are present, but distributed over many different systems. The distribution of the data makes it difficult to apply CBR in real-time, as data have to be collected from the different systems. In this work we propose a framework and algorithm to efficiently build a case representation on-demand and solve the challenge of distributed data in CBR. The main contribution of this work is a framework using an index for objects and the sources where data about those objects can be found. Next to the framework, we present an algorithm that operates on the framework and can be used to build case representations and construct a case base on-demand, using data from distributed sources. There are several parameters that influence the performance of the framework. Accordingly, we show in a conceptual and experimental evaluation that in highly-distributed and segregated environments the proposed approach reduces the time complexity from polynomial to linear order.