1 A Method for Matching Crowd-sourced and Authoritative Geospatial Data Heshan Du, Natasha Alechina, Michael Jackson, Glen Hart University of Nottingham, UK Abstract A method for matching crowd-sourced and authoritative geospatial data is presented. A level of tolerance is defined as an input parameter as some difference in the geometry representation of a spatial object is to be expected. The method generates matches between spatial objects using location information and lexical information, such as names and types, and verifies consistency of matches using reasoning in qualitative spatial logic and description logic. We test the method by matching geospatial data from OpenStreetMap and the national mapping agencies of Great Britain and France. We also analyze how the level of tolerance affects the precision and recall of matching results for the same geographic area using 12 different levels of tolerance within a range of 1 to 80 meters. The generated matches show potential in helping enrich and update geospatial data. 1. Introduction Maps, whether digital or paper-based, are a common feature of our daily life. They typically provide a two-dimensional representation of geographic features, such as roads, rivers, buildings, places, etc., in the real world (i.e. a topographic base) over which other ‘thematic’ information may be displayed such as density of population or crime statistics. The information represented provides both an indication of where on the earth’s surface an object of interest is (i.e. its geometry) and lexical information on what that geometry represents (e.g. a road and its name such as ‘High Street’). Such information represented in maps is often referred to as geospatial data and plays an essential role in many governmental, economic and social operations, such as disaster response, urban planning and tourism. Traditionally, most national level mapping was carried-out by government agencies or specialist mapping companies, because it required the use of expensive or difficult-to-obtain survey data, plus specialist tools and later software and an associated high-level of expertise. Geospatial data which is surveyed and classified using formal quality assurance procedures, for example by a national mapping agency, is referred to be ‘authoritative’. Maps produced by the general public, who did not have access to such data sources, nor the specialist tools and software, focused more on smaller areas and on indicating where key features were in relative terms but typically could not be relied upon for precise location, completeness or consistency. This situation has been radically changed in recent years by a number of technological developments and by governments through the release of associated data (e.g. precise Global Navigation Satellite System data, satellite and aerial imagery). Perhaps the most important of these developments is the mobile smartphone. Such phones are capable of accurately recording their positions and com- bined with the use of simple-to-use applications can delimit physical and man-made features and tag the resulting geometries with information describing the nature, purpose and use of those features. This ‘crowd-sourced data’ may be actively collected as a volunteer activity by citizens (Goodchild 2007) or passively acquired as a bi-product of an application the main purpose of