Mining Event-Related Knowledge from OpenStreetMap

Abstract

With an explosive growth in the number of contributors for creating and assembling of spatial data, freely available databases and open source products have drawn the attention among decision makers for facility management and service planning. Many location-based services are using Volunteered Geographic Information (VGI) as spatial data sources. The key motivation of this work is to mine hidden patterns of social activities and the interests of contributors to share event-related knowledge within OSM community as one of the most prominent examples of user generated spatial data. In this study, the term event referred to anomalous user activities, number of contributors plus number of contributions, which happened at a time point or within a specific period of time. We focused on events which have happened and the events for which we had prior knowledge. For the purpose of retrospective event detection, it is necessary to analyse the history of OSM for the area of the event. In our case, the entire OSM history of Vechta, Munich, Los Angeles, and Sendai, around the area where the events happened was extracted to examine the potential of OSM database for event detection. Our experimental analysis reveals that while changes to OSM can be effectively rendered on the globally visible OSM maps in a few hours, citizens would not naturally use OSM as a tool to mark an event. In fact, the contributors do not treat this community in the same way as they do with other user-friendly electronic exchange platforms such as Twitter, Face book, or Flickr. The obtained results also show that for big events such as tsunami in Sendai during which the geometry of objects is affected, post-disaster, structural and environmental damages (demolished buildings, road infrastructure changes, etc.) are detectable through OSM.KeywordsEvent detectionVolunteered geographic information (VGI)OpenStreetMap (OSM)Location-based services (LBS)Web 2.0