Quadtree-based Resource Description Techniques for Spatial Data in Distributed Databases

Abstract

In the social media age, content creation and distribution of all sorts of digital media is of growing importance. The variety of media types, devices, and user groups results in the generated media items being stored in a very heterogeneous landscape which can be seen as a large distributed database consisting of personal computers, mobile devices, or web servers of social media sites, demanding for an appropriate overarching search system. Resource selection based on compact resource descriptions allows to e ciently determine resources maintaining relevant media items. Among other aspects, geospatial footprints have to be addressed for an e ective media search. We propose and evaluate quadtree-based techniques to summarize collections of geo-referenced media items in a distributed geographic IR context. Based on these summaries, resource selection decisions are made when searching for media items close to a given geographic location. The quadtree-based approaches are evaluated against previous work. The quadtree-based summaries prove to be very flexible and competetive.