Quality Management of Volunteered Geographic Information: Quality Assessment and Representation

A growing number of cell phones, digital cameras, PDAs and other hand-held devices are equipped with georeferenced data collection technologies, made it possible for ordinary people to collect spatial data, which are then shared and disseminated on the internet using web map services. It has led to a huge source of spatial data termed as Volunteered Geographic Information(VGI) by Mike Goodchild. Volunteered geographic data are constantly being added, edited or removed by the users, so different versions of the same data may exist. It is very common in geographic communities to confront with several versions of the same data and select the relevant one based on the quality parameters stored in the metadata. However, in case of VGI, the users are not experts and do not necessarily have high spatial knowledge. Therefore, a system administrator decides on behalf of the users to select and publish one of the produced data as the best, which is in most cases the latest one, because they believe that if the data is not correct, someone will improve it later. However, this is not the right assumption. Quality of spatial data depends on different parameters such as spatial accuracy, attribute accuracy, completeness, logical consistency and updateness. Therefore, the best data may differ from an application to another.In this article, we propose providing the VGI users with the spatial data quality parameters through simple cartographic representation and let them decide on selecting the appropriate dataset for the application in hand. The users select the desired quality parameters as well as the visual element (e.g. color, intensity, style, thickness, etc.) to classify the desired parameters. All the datasets are represented by the selected visual element based on their metadata information, which help the users to visually compare them and select their appropriate dataset. In addition to determine the quality of each data set, instead of comparing them with the ground truth data they’ll be compared with each other to define the relative quality of them.The proposed approach has been implemented for a case study where 16 versions of 2D map were produced using different data collection methods (i.e., surveying, GPS and digitization) and with different spatial data quality parameters, limited here to spatial accuracy and completeness. An ArcGIS extension was developed to display the datasets based on the visual classification elements as well as spatial quality parameters selected by the user, which eventually help the user to visually compare the existing data and select the appropriate one for his/her certain application.