Evaluating a thesaurus for discovery of ecological data

The increasing availability of data has driven a need for improved capabilities to discover data. Use of keywords drawn from a thesaurus is one way to enable browse-based discovery and to enhance searching. To assess the effect of using a thesaurus on the discoverability of ecological data, use of 81,415 keywords derived from 6132 data packages drawn from 28 ecological research projects in the U.S. Long-term Ecological Research Network were examined. The vast majority (95%) of data packages included at least one keyword drawn from the thesaurus, thus enabling their discovery using a hierarchical browse interface. For searching, keywords derived from the thesaurus would reveal 17 times more data packages than ad hoc keywords not in the thesaurus. Additionally, searches using keywords derived from the thesaurus returned data from a median of four different projects, whereas ad hoc search terms would typically yield data from only a single project. Of the search terms that yielded more than five data packages across two or more projects, 78% were found in the thesaurus. Use of keywords drawn from the thesaurus increased when compared to their use prior to establishment of the thesaurus, indicating that terms from the thesaurus are being actively added to metadata. A questionnaire assessed the process by which keywords were selected and indicated that information management personnel played an important role in assigning keywords drawn from the thesaurus. These results support the idea that adoption of a thesaurus can be an effective way to enhance the discoverability of ecological data, and that keywording practices play an important role in supporting that enhancement.