Enriching and validating geographic information on the web

Nicolas Tempelmeier, Technische Informationsbibliothek (TIB)
2022
The continuous growth of available data on the World Wide Web has led to an unprecedented amount of available information. However, the enormous variance in data quality and trustworthiness of information sources impairs the great potential of the large amount of vacant information. This observation especially applies to geographic information on the Web, i.e., information describing entities that are located on the Earth's surface. With the advent of mobile devices, the impact of geographic
more » ... information on our everyday life has substantially grown. The mobile devices have also enabled the creation of novel data sources such as OpenStreetMap (OSM), a collaborative crowd-sourced map providing open cartographic information. Today, we use geographic information in many applications, including routing, location recommendation, or geographic question answering. The processing of geographic Web information yields unique challenges. First, the descriptions of geographic entities on the Web are typically not validated. Since not all Web information sources are trustworthy, the correctness of some geographic Web entities is questionable. Second, geographic information sources on the Web are typically isolated from each other. The missing integration of information sources hinders the efficient use of geographic Web information for many applications. Third, the description of geographic entities is typically incomplete. Depending on the application, missing information is a decisive criterion for (not) using a particular data source. Due to the large scale of the Web, the manual correction of these problems is usually not feasible such that automated approaches are required. In this thesis, we tackle these challenges from three different angles. (i) Validation of geographic Web information: We validate geographic Web information by detecting vandalism in OpenStreetMap, for instance, the replacement of a street name with advertisement. To this end, we present the OVID model for automated vandalism detection in OpenStreet [...]
doi:10.15488/12283 fatcat:6wur6svrs5bohdibdmofakaebe