Using Geographic Relevance (GR) to contextualize structured and unstructured spatial data

Geographic relevance is a concept that has been used to improve spatial information retrieval on mobile devices, but the idea of geographic relevance has several potential applications outside of mobile computing. Geographic relevance is used measure how related two spatial entities are using a set of criteria such as distance between features, the semantic similarity of feature names or clustering pattern of features. This thesis examines the use of geographic relevance to organize and filter web based spatial data such as framework data from open data portals and unstructured volunteer geographic information generated from social media or map-based surveys. There are many new users and producers of geographic information and it is unclear to new users which data sets they should use to solve a given problem. Governments and organizations also have access to a growing volume of volunteer geographic information but current models for matching citizen generated information to locations of concern to support filtering and reporting are inadequate. For both problems, there is an opportunity to develop semi-automated solutions using geographic relevance metrics such as topicality, spatial proximity, cluster and co-location. In this thesis, two geographic relevance models were developed using Python and PostgreSQL to measure relevance and identify relationships between structured framework data and unstructured VGI in order to support data organization, retrieval and filtering. This idea was explored through two related case studies and prototype applications. The first study developed a prototype application to retrieve spatial data from open data portals using four geographic relevance criteria which included topicality, proximity, co-location and cluster co-location. The second study developed a prototype application that matches VGI data to authoritative framework data to dynamically summarize and organize unstructured VGI data. This thesis demonstrates two possible approaches for using GR metrics to evaluate spatial relevance between large data sets and individual features. This thesis evaluates the effectiveness of GR metrics for performing spatial relevance analysis and it demonstrates two potential use cases for GR