Sensor Web Recommender System Harnessing collaborative intelligence for geospatial information discovery.

摘要

This thesis reports on the research undertaken to build a geospatial context-aware recommender system for the Sensor Web. The recommender system serves to enhance information filtering and knowledge discovery capabilities for users of the Sensor Web. The sensors, the sensor observations, and the explicit user feedback all contain geospatial information.;In addition to the research on the two components mentioned above, our research also focused on the question of visualizing large folksonomies as dynamic real-time tag maps. This thesis reports on a solution to that problem.;The recommender system leverages that location information to extend traditional recommender system theory to geospatial contexts. The recommender system captures both evaluative and descriptive feedback as ratings and tags respectively. Thus, the recommender system contains two distinct components: the folksonomy engine and the rating engine. We present the theoretical underpinnings, describe the implementation, and discuss the results of the evaluation of each component.