Tackling the thematic accuracy of areal features in OpenStreetMap

With the increasing importance of VGI for GIScience, data quality becomes an issue of high concern. Particularly in collaborative mapping projects, when a group of public participants acts to collect, update and share information about geographic features, aiming to maintain and improve a geo-spatial dataset. OpenStreetMap (OSM) is the most common VGI project that aims to develop free world digital map. Although several studies emphasized the positional accuracy and completeness of the OSM data, particularly in the urban areas, they also highlighted its problematic thematic accuracy. In this chapter, we handle the thematic accuracy quality measure from the facet of classification. This chapter presents an approach for rule-guided classification for VGI projects. The proposed approach exploits the availability of data to learn the distinct characteristics of a set of geographic features. Afterwards, the learned characteristics are used to guide the contributors toward the most appropriate data classes, aiming to improve the data quality. The approach consists of two phases: Learning and Guiding phases. During the Learning phase, data mining algorithms are applied to learn the geographic characteristics of specific features. The learning process results in a set of rules describing these features. The extracted rules are used to develop a classifier. Afterwards, during the Guiding phase, the developed classifier is used for several purposes; 1) acts to detect problematic classified entities; and 2) guides and aids the contributors during the classification process. An empirical study followed by an implementation is conducted. The results show the feasibility of the proposed approach and highlight some limitations that could be improved in the future studies. The developed tool generates promising results and improves the classification of OSM dataset as well.