Semi-Automated Visualization of Spatial Context in Unstructured Text
ISBN 978-85-88783-11-9
Authors
1Chiang, Y.; 2Gehring, S.
1SPATIAL SCIENCES INSTITUTE,UNIVERSITY OF SOUTHERN CALIFORNIA Email: yaoyic@usc.edu
2SPATIAL SCIENCES INSTITUTE,UNIVERSITY OF SOUTHERN CALIFORNIA Email: sgehring@usc.edu
Abstract
Digital information with a spatial component is being generated at an astounding rate, from sources such as Flickr Videos, online news, and “tweets” on Twitter. The ability to identify locations in unstructured text and quickly generate a map unlocks valuable information about the context of locations in the text. Geoparsing, the process of assigning geographic coordinates or other geographic identifiers to unstructured text, extracts this valuable information from text. Existing studies and tools focus on the challenges of location extraction and disambiguation. These studies do not focus on visualizing the extracted locations, and generally use a simple method of displaying each location as a single point on a map. This paper examines the current geopars-ing text-to-map applications, identifies challenges to generate a map from a text document, and defines an approach to display locations with their spatial context (e.g., administrative boundaries) on a map. The outcome of this paper is a semi-automated geoparsing, data integration, and visualization approach to convert the locations in text-based news articles to locations with contextual information on a map. This approach provides an efficient and effective way to display the spatial context of a text document and allow for interpretations of the data that is not readily apparent from the text by itself.