Richard M. Keller, Rajkumar Thirumalainambi, Joseph C. Coughlan
One of the main barriers to exploiting the great wealth of global earth science data available today is that researchers are unable to rapidly search and find data relevant to their studies. This data is spread across a large number of archives maintained by different institutions employing a bewildering array of different data description languages. In this paper, we describe a metadata federation approach designed to support queries across multiple earth science data archives without requiring the adoption of a unified metadata standard. Our ontology-based approach employs a central metadata transformation facility capable of integrating heterogeneous metadata using a set of translators and wrappers. This shifts the burden of federation from the data provider to the central metadata facility, acknowledging that not all data providers have the motivation or resources to comply with externally-imposed metadata standards. We demonstrate the feasibility of this approach with a proof-of-concept prototype that federates metadata across two earth science data archives - one containing NASA data and the other containing USDA data — despite the differences in their metadata languages.
Subjects: 1.6 Engineering And Science; 11.2 Ontologies
Submitted: Jan 24, 2008