Modeling Accessibility through Geocrowdsourcing



Qin, Han

Journal Title

Journal ISSN

Volume Title



Map-based crowdsourcing is one of the most significant contemporary trends in the geospatial sciences and has completely changed many data collection workflows, and added new sources of data. An important aspect of this emerging trend is the manner in which data quality is assessed, and how well these quality assessment processes match processes used in traditional map-based and geographic information systems-based quality assessment procedures. This dissertation studies the evolution of geographic data collection, and the methods of quality assessment, and builds a comprehensive quality assessment workflow for geocrowdsourced data. This workflow is based on many traditional formulations of quality, such as positional accuracy, temporal consistency, categorical accuracy, fitness-for-use, and lineage. These quality assessment workflows are studied through the George Mason University Geocrowdsourcing Testbed (GMU-GcT), which was designed to study dynamic aspects of map-based crowdsourcing. The GMU-GcT tests the implementation of techniques from the US National Map Accuracy Standard (NMAS) as well as the National Standard for Spatial Data Accuracy (NSSDA), as well as several new techniques, modified over time, that are shown to have value within the specific context of geocrowdsourcing conducted with the GMU-GcT. This research extends the quality assessment work with modeling of a pedestrian network and the accessibility characteristics associated with navigation obstacles, many of which have been crowdsourced with the GMU-GcT, and tests the feasibility of infrastructure maintenance using geocrowdsourced data and associated quality assessment parameters. The quality assessment techniques from traditional mapping domains are shown to have value in the domain of geocrowdsourcing, and the ability to model pedestrian network accessibility and maintenance optimization is demonstrated through this work. Extensions of this research into geosocial media is explored with mixed results, and future work in simplified, image-based geocrowdsourcing is explored to determine what quality assessment metrics can be derived from greatly simplified geocrowdsourcing methods. Additional modeling enhancements, based on alternative optimization strategies and weighting factors, is discussed as a future area for work. Summary of end-user and subject matter experts is discussed in context of future modifications to the GMU-GcT.



Accessibility, Data Quality, Geocrowdsourcing, Optimization Modeling