ScienceBase: MAKING CONNECTIONS WITH METADATA INTEGRATION

ScienceBase is a U.S. Geological Survey effort to help better manage scientific data and information. ScienceBase is both n aggregation of catalogs that provides fresh exposure to different collections of scientific data and a new, original repository for both data and metadata. ScienceBase addresses the problem of not only getting to scientific data resources but to track back through the chain of projects where the data originated, scientists involved in collection, and manuscripts that discuss findings.

In the process of developing ScienceBase from the grass roots in partnership with science teams a number of interesting dynamics have arisen. ScienceBase makes connections between potentially related items when harvesting from various other catalogs and facilitates users making connections when new items are uploaded and documented. The integration steps involve the following elements at various stages of maturity:

• Contacts - authors and other contacts associated with items
• Places - geospatial proximity or similar geotags
• Keywords - use of terms in descriptive information
• Projects - connection to a record describing a science project
• Citations - relationship to a record describing a publication or other citation

Each of these present a variety of integration challenges, demanding approaches that use both standards and conventions along with fuzzy logic in determining proximal relationships between two items. When items are harvested from different and usually unrelated sources, the information about the items is taken out of one context and put into another. This re-contextualization process can result in a loss of original meaning when the metadata describing an item are reliant on the originating context.

ScienceBase computer and information science work is attempting to navigate these issues through a variety of different techniques and methods, striving to connect scientists with an ever improving set of links to more information in the same way that consumers are linked to potential products of interest. Work involves both explicit means with linked data techniques, research into implicit methods involving usage analysis, and aspirations toward a higher degree of formalized user knowledge capture where scientists using ScienceBase can help connect the dots through their own actions.