A Parallel Architecture Model for Data-Driven Conceptual Clustering Methods

摘要

For years now we have been devoting a lot of attention to the design of conceptual clustering algorithms able to process very large corpora of data, as requested by large scale data mining applications. Considering the development of parallel computers, we decided to transpose one of our algorithms, called MSG, on a MIMD parallel computer. The MSG is an instantiation of a universal representation paradigm for conceptual clustering methods. By mapping it to a parallel architecture, we hope to provide insights on how to use such architectures to push the limits of these methods further in terms of the volume of data that they can handle.