AbstractThis thesis arose from the recognised lack of previous research into the usability of geospatial data. Whereas considerable study has been made into the usability of devices and their interface, the underlying data has been subject to less consideration, though a substantial literature exists regarding data quality. With the rapid expansion of geospatial data availability through a multitude of platforms (much of it free and crowdsourced), and the increase in creation of such data as mobile devices track location, there is a need to investigate the usability of the data in different contexts and applications. This thesis contends that data quality and data usability, though closely related, are separate characteristics, and that quality is an important element of data usability.
Usability of different data types from various sources is examined here in the context of the well-established application of GIS-based accessibility modelling. Sensitivity analysis techniques were utilised in a novel way to highlight usability issues with the data being studied through the use of statistical and visual approaches. Comparisons were made between a variety of proprietary datasets and data from other sources, such as free and open-source software (FOSS) volunteer geographic information (VGI) network data from OpenStreetMap (OSM), and observations made as to their usability, while addressing cross-cutting topical themes, such as examining different sources of locational representation, different sources of network representation, and questioning the effect of supply and demand on accessibility. The use of both an urban and a rural study area enabled comparisons to be drawn in different geographical contexts.
Several specific proposals are made with regard to improving usability of the proprietary (Ordnance Survey) and VGI (OSM) datasets. An aide to decision making is also suggested through the use of a usability checklist, enabling sample or trial data to be assessed quickly and simply in any given context. A novel Utility Factor is proposed, which draws together contributions from the quantitative aspects of this study into one figure, and is suggested as a context-based proxy of usability based on measures of data similarity, difference and effect.
The results obtained confirm the need for further research to both clarify aspects of data usability and widen the scope of future research into different contexts and applications.
|Date of Award
|Gary Higgs (Supervisor), Mark Ware (Supervisor) & Mitchel Langford (Supervisor)