Model-Based Information Integration in a Neuroscience Mediator System

We present the information mediator prototype called Kind [GLM00], recently developed as part of an integrated Neuroscience workbench project at SDSC/UCSD within the NPACI project. The broad goal of the workbench is to serve as an environment where, among other tasks, the Neuroscientist can query a mediator to retrieve information from across a number of information sources, and use the results to perform her own analysis on the data. The Kind mediator is an instance of a novel model-centered mediator architecture that extends current XML-based mediator approaches by incorporating a semantic model of an information source as an integral part of the mediation process. Thus, by model we mean a combination of (i) a conceptual model of the source data, for example a UML or EER model, and (ii) any addi-