Strategies for geographical scoping and improving a gazetteer

Abstract

Many applications that use geographical databases (a.k.a. gazetteers) rely on the accuracy of the information in the database. However, poor data quality is an issue when data is integrated from multiple sources with different quality constraints and sometimes with little information about the sources. One major consequence of this is that the geographical scope of a location and/or its position may not be known or may not be accurate. In this paper, we study the problem of detecting the scope of locations in a geographical database and its applications in identifying inconsistencies and improving the quality of a gazetteer. We develop novel strategies, including probabilistic and geometric approaches, to accurately derive the geographical scope of places based on the spatial hierarchy of a gazetteer as well as other public information (such as area) that may be available. We show how the boundary information derived here can be useful in identifying inconsistencies, enhancing the location hierarchy and improving the applications that rely on gazetteers. Our experimental evaluation on two public-domain gazetteers reveals that the proposed approaches significantly outperform, in terms of the accuracy of the geographical bounding boxes, a baseline that is based on the parent-child relationship of a gazetteer. Among applications, we show that the boundary information derived here can move more than 20% of locations in a public gazetteer to better positions in the hierarchy and that the accuracy of those moves is over 90%.