An E-commerce System for Mining Distributed Spatial Databases

-Due to restr ictive data access and a lack of appropr iate data mining software, spatial information from physically dispersed sites is often not proper ly exploited in ecommerce. In the proposed distr ibuted spatial knowledge discovery system for e-commerce, a secure centralized server collects propr ietary heterogeneous data from subscribed businesses as well as relevant data from public and commercial sources and then integrates knowledge to provide valuable management information to subscr ibers. Considered knowledge discovery methods include: (1) providing estimated values for unobserved, typically expensive attr ibutes of interest to a par ticular business; or (2) deliver ing learned models for generalizing extracted knowledge. An evaluation on large, highly nonlinear simulated data suggests that both approaches can provide profitable, effective and useful management recommendations in spatial e-commerce applications.