Dr. Wenwen Li
GeoDa Center for Geospatial Analysis and Computaion School of Geographical Sciences and Urban Planning
Polar studies have attracted much attention in recent years because this region is of great importance to our planet for being the key drivers of the Earth climate, the sources of rich mineral resources, and the home of a variety of marine biology. However, the global warming over the past century is pushing the polar systems to reach a tipping point, at which the systems encounter high risk from the melting of snow and sea ice covers, permafrost thawing, and acidification of the Arctic oceans. To increase the understanding of the polar environment, the National Science Foundation has established a Polar Cyberinfrastructure (CI) program, aiming at utilizing the advanced software architecture – the cyberinfrastructure to support polar data analysis and decision-making. At the center of this Polar CI research is data and the link between scientists and the most needed dataset. This paper reports our efforts in developing a semantic search tool that supports the intelligent discovery of polar datasets. This tool is built upon latent semantic analysis technique and improves the search performance through the identification of hidden semantic associations between terminologies used in the metadata of the various datasets. The software tool is implemented using a model-view-control (MVC) design pattern and has been successfully integrated into a very popular open source metadata catalog as a new semantic search support. A semantic matrix is maintained persistently within the catalogue to store the semantic associations and a dynamic update mechanism is developed to allow automated update of semantics once more metadata are loaded into or removed from the catalog. We explored the effects of stemming and rank reduction to the effectiveness of this semantic search module and demonstrated its better performance than the traditional search techniques.
Wednesday March 12, 2014 10:00am
Refreshments Served: 9:45am
FL2-1001 (Small Seminar Room)