Spatial bayesian learning algorithms for geographic information retrieval
Proceedings of the 13th annual ACM international workshop on Geographic information systems
SCOUT: prefetching for latent structure following queries
Proceedings of the VLDB Endowment
| Hi-index | 0.00 |
Existing Geographic Information Systems (GIS) areintended for expert users and consequently, do notprovide any machine intelligence to assist users. Thispaper presents a Bayesian framework that willincorporate expert knowledge in order to retrieve allrelevant datasets given an initial user query. Theframework uses a spatial model that combines relational,non-spatial and spatial data. This spatial model allowsefficient access of relational linkages for a Bayesiannetwork, and thus improves support for complex andvague queries. The Bayesian network assigns causalprobabilities to these relational linkages in order todefine expert knowledge of related datasets in the GIS. Inaddition, the framework will learn which datasets arebest suited for particular query input through feedbacksupplied by the user.This contribution will increase the performance andefficiency of knowledge extraction from GIS by allowingusers to focus on interpreting data, instead of focusing onfinding which data is relevant to their analysis. Theinitial user query can be vague and the framework willstill be capable of retrieving relevant datasets via thelinkages discovered in the Bayesian network.