Files
Abstract
Place is an abstract concept which is difficult to describe or define in a comprehensive and objective manner. In terms of data construction, there is a wide gap between how people describe places in daily life and places represented in databases. Traditionally, places are recorded by authoritative agencies using a simple structure with the dataset of place name, its type, and footprints. For the last few years, as platial data gets more attention in the field of GIS (Geographic Information Systems), there is a need to represent knowledge about places more specifically in terms of data semantics. This dissertation aims to propose an organizational architecture of place concepts that combines elements from multiple data sources using a computational ontological approach. It is important to describe how traditional and non-traditional data sources can be used to represent place information in an ontology. By comparing different concepts and the relations used in different sources, this study intends to outline the consistent components that can be used in a place ontology. To achieve this, traditional gazetteers such as the Geographic Names Information System (GNIS), GeoNames, the Getty Thesaurus of Geographic Names (TGN) and non-traditional data sources such as Google Places and Twitter data were examined to identify the most often utilized concepts and relations when demarcating place information. As a pilot study, Twitter data was used to explore people’s activities to better understand how people utilize places and to assess the semantics among place categories.
Finally, place information extracted from various data sources is represented in ontological syntax. GeoSPARQL standard developed by the Open Geospatial Consortium (OGC) was reused in the ontology to describe geometries and the properties of spatial relations among places. Findings in this study, including a summary of comparisons with existing gazetteer ontologies, demonstrate the potential for using expert and non-expert (e.g., social media users) knowledge to enrich place knowledge. Thus, this dissertation contributes to discourses on digital gazetteers and geographic information retrieval (GIR). The overall framework of this study can be applied to the GIS web to increase information accessibility about places with a wide range of semantics.