An exposition of the nature of volunteered geographical information and its suitability for integration into spatial data infrastructures

Abstract

This thesis presents an analysis of the nature of volunteered geographical information (VGI) and on its applicability for use in a spatial data infrastructure (SDI) to supplement official and commercial sources, particularly given the ease with which ordinary people can document their environment, experiences, perspectives and prejudices, share them widely and rapidly, and even query anyone else?s content. For this research, taxonomies and repositories of such information were examined qualitatively and using formal concept analysis (FCA). Further, this thesis attempts to reflect on the context for SDIs and VGI and the challenges and opportunities for both. An SDI is an evolving concept for facilitating and coordinating the management and sharing of geospatial data, with services, metadata, products, standards and inter-organisation arrangements and structures. It can take long to establish an SDI; some have failed and they have competition. In South Africa, the National Development Plan has an objective to establish a national spatial observatory: it is part of an SDI with its own value-add data, and products provided through the SDI or directly to its participants. The Spatial Data Infrastructure Act established the South African Spatial Data Infrastructure and its Committee for Spatial Information. Creating vast quantities of user-generated content (UGC) has been enabled by the pervasiveness, power and affordability of inter-networking, social media, virtual communities, applications and mobile devices. VGI is user-generated content with geospatial components, or user-generated geospatial content. VGI can contribute successfully to an SDI, at the local, national, regional or global level. VGI can extend the reach in time and space of official mapping agencies and others contributing to an SDI, because of the sheer volume of humans and their devices acting together or independently, as sensors, recorders and disseminators. VGI; repositories of VGI; innovative integration of content, applications and services (mashups); crowd sourcing and new geographical theories (psychogeography, social theory, social justice, ethics, etc) all challenge the traditional business models of SDIs. However, metadata, quality, classification and standards can be challenges for VGI. Further, while some VGI can be useful, other VGI can be spurious, misleading or wrong. There are also different interpretations over what is actually VGI. To provide context for the exposition, this thesis also examines terminology, geospatial data, classification, folksonomies, virtual globes, inter-networking, the limitations of the Internet, controlling the Internet, privacy, exploiting content, social media, curation, the digital divide, citizen science, crowd sourcing, neogeography, metadata, quality, standards and formal concept analysis (FCA). To determine the nature of VGI and its suitability for an SDI, this thesis investigates various taxonomies of UGC, VGI and citizen science; assesses qualitatively their discrimination adequacy using VGI repositories; and assesses them using FCA. This thesis also presents original research contributions, to information science, geographical information science and theoretical computer science. For FCA it presents lemmas on stability in a lattice (providing lower and upper bounds for intensional and extensional stability indices), it shows there is value in instability in a lattice when assessing a taxonomy (representing extreme values rather than noise) and it presents stability exploration, a possible decision support tool. It describes the four stages for recognising the quality of a resource, it reports on a survey of geographical information professionals on VGI, SDIs and virtual globes, and it clarifies the differences between UGC, VGI, citizen science, crowd sourcing and neogeography, which can be confused with one another. Finally, this thesis explains why the Internet cannot be controlled.