Distributed Knowledge Map for Mining Data on Grid Platforms

Recently, huge datasets representing different applications domains are produced and stored on distributed platforms. These datasets are, generally, owned by different organizations. As a consequence, The scale and distribution nature of these datasets have created the problem of efficient mining and management on these platforms. Most of the existing knowledge management approaches are mainly for centralized data mining. Few of them propose solutions for mining and handling knowledge on Grid. However, the new knowledge is stored and managed as any other kinds of resources. To solve this problem, we introduce a "distributed knowledge map", which represents easily and efficiently the new knowledge mined from these very large distributed platforms such as Grids. This approach is developed and integrated as part of our distributed data mining framework. This knowledge map also facilitates the integration/coordination of local mining processes and existing knowledge to increase the accuracy of the final model. Our knowledge map is tested on real large datasets.