AbstractsGeography &GIS

In recent years Crowdsourced, or Volunteered, Geographic Information (CGI, VGI), has emerged as a large, up-to-date and easily accessible data source. Primarily attributable to the rise of the Geoweb and widespread use of location enabled technologies, this environment of widespread innovation has repositioned the role of consumers of spatial information. Collaborative and participatory web environments have led to a democratisation of the global mapping process, and resulted in a paradigm shift to the consumer of geographic data also acting as a data producer. With such a large and diverse group of participants actively mapping the globe, the resulting flood of information has become increasingly attractive to authoritative mapping agencies, in order to augment their own spatial data supply chains. The use of CGI would allow these agencies to undertake continuous improvement of their own data and products, adding a dimension of currency that has previously been unattainable due to high associated costs. CGI, however, through its diversity of authorship, presents a quality assurance risk to these agencies should it be included in their authoritative products. Until now, this risk has been insurmountable, with CGI remaining a “Pandora’s Box” which many agencies are reluctant to open. This research presents an algorithmic model that overcomes these issues, by quantifying trust in CGI in order to assess its implied quality. Labeled “VGTrust”, this model assesses information about a data author, its spatial trust, as well as its temporal trust, in order to produce an overall metric that is easy to understand and interpret. The VGTrust model will allow mapping agencies to harness CGI to augment existing datasets, or create new ones, thereby facilitating a targeted quality assurance process and minimizing risk to authoritativeness. This research proposes VGTrust in theory, on the basis of existing examinations of trust issues with CGI. Furthermore, a facilitated case study, “Building Our Footprints” is presented, where VGTrust is deployed to facilitate the capture of a building footprint dataset, the results of which revealing the veracity of the model as a measure to assess trust for these data. Finally, a data structure is proposed in the form of a “geo-molecule”, which allows the full spectrum of trust indicators to be stored a data structure at feature level, allowing the transitivity of this information to travel with each feature following creation. By overcoming the trust issues inherent in CGI, this research will allow the integration of crowdsourced and authoritative data, thereby leveraging the power of the crowd for productive and innovative re-use.