InfoFuse: Interleaved Information Gathering and Reasoning for Information Fusion

Abstract

Abstract : The vast amount of geospatial data now available covering the entire world presents new and exciting opportunities to derive new information through information fusion. These data sources include mapping services (Google Maps. Yahoo Maps. etc.). Web 2.0 based collaborative projects (WikiMapia and OpenStreetMap). traditional geospatial data sources (raster maps, KML vector layers), and non-traditional geospatial data sources (phone books, property records, etc.). This large amount of diverse data increases the probability of encountering missing or inconsistent data and requires efficient reasoning algorithms to scale to large problem instances during information fusion. To address these issues, we have developed a geospatial fusion framework that integrates the various types of geospatial data available within a region. Our approach builds on our past work on constraint satisfaction reasoning and data access. This framework supports the ability to gather and fuse information, and uses conflict resolution strategies to disambiguate data inconsistencies. We implemented our approach into a system called InfoFuse and successfully demonstrate this approach on the real-world data for Belgrade.