GEOVISUALIZATION FOR THE SPATIAL, TEMPORAL AND SEMANTIC CONTEXT OF NEWS

A. Villacorta1, J. Dillemuth2

1 - University of California, Santa Barbara, Dept. of Statistics and Applied Probability, Santa Barbara, California, USA

2 - University of California, Santa Barbara, Dept. of Geography, Santa Barbara, California, USA

villacorta@umail.ucsb.edu, julie@geog.ucsb.edu

 

News events may be considered to occur in each of three subspaces: temporal, geographic, and semantic space. An event occurs at a specific location over a range of time, and news about the event spreads, over time, to other locations. But a query to a typical database of news archives returns only a list of articles, with limited means of helping the user understand the temporal and geographic context of the news story as a whole. These temporal and geospatial components of news stories are significant in analyses of information movement: when and where did an event happen, and when and where was the story published? Using a case study of the Terri Schiavo story from 2005, we present a framework for how text-based spatiotemporal and semantic information from a database search can be transformed into geovisualizations, facilitating observation, analysis, and communication. Current news visualizers, which typically use the Google News aggregator as their data source, focus on either the spatial or temporal component, but not both. For a student or analyst, understanding the full context of a news story allows him or her to potentially detect patterns and make new insights.

Our case study on Terri Schiavo news coverage is an example of a local story that quickly gained national attention, and almost as quickly faded from the medias coverage. Terri Schiavo was a Florida woman in a coma whose family was involved in a legal battle over her right to die. We consider all the news articles returned by a keyword search in the Americas Newspapers (Newsbank, Inc.) digital archive on Terri/y Schiavo, limiting our spatial domain to two states, California and Texas. We present multiple geovisualizations based on frequency counts of reporting, circulation statistics, population density, time lags between reports, and a semantic analysis of the headlines using Latent Semantic Indexing and Self Organizing Maps. The goal of these various cartographic approaches is to explore and communicate the patterns and context that emerge from looking at data from different perspectives. To highlight the relevance of search results, we also classify the articles as either about the events themselves, tangentially-related, or unrelated. Our results demonstrate the value of understanding the spatiotemporal context of news, and ways to achieve that understanding through geovisualization.