Authors: Chuanrong Zhang*, University of Connecticut, Department of Geography, Tian Zhao, Department of Computer Science, University of Wisconsin-Milwaukee, Weidong Li, aDepartment of Geography & Center for Environmental Sciences and Engineering, University of Connecticut
Topics: Geographic Information Science and Systems
Keywords: Geospatial Semantic Web, Optimization, GeoSPARQL, RDF, Query performance
Session Type: Paper
Start / End Time: 8:00 AM / 9:40 AM
Room: Roosevelt 3, Marriott, Exhibition Level
Presentation File: No File Uploaded
The GeoSPARQL was developed for Querying RDF data on Geospatial Semantic Web. However, GeoSPAQL query may become very slow when query a large number of spatial objects. This study investigated several optimization techniques used for improving performance of GeoSPARQL query. We focused on the client side optimization techniques in this study. The major investigated techniques include on-the-fly spatial indexing, tile-based rendering, query rewriting algorithms, efficient algorithms for spatial join and cache techniques. We conducted some experiments on a client computer for performance evaluation of using these optimization techniques. Our experimental results show that the Geospatial Semantic Web query suffered from the slow performance without using these optimization techniques. The optimization techniques such as on-the-fly spatial index, tile-based rendering, and cache techniques can greatly improve the performance of GeoSPARQL query. The spatial join queries using different spatial join algorithms such as nested-loop algorithm, R-Tree index nested-loop algorithm, and tiled-based rendering algorithm have different performance. Although these optimization techniques can improve the performance of the prototype, there are still some limitations. Other optimization techniques such as cloud and parallel computing, or GPU techniques should be adopted in the future to further improve the Geospatial Semantic Web query.