Home > Demo
This is a demo of using emphatic patterns of word usage learned from a large number of documents to automatically identify the perspective from which a text is written. We develop a statistical model called a Joint Topic and Perspective Model to capture emphatic patterns and infer the perspective of a document. The first version focuses on the Israeli-Palestinian conflict, and uses the patterns learned from the bitterlemons corpus.
This is a demo of automatic discovering emphatic patterns of word usage from texts that are written or spoken from contrasting ideological perspectives. The joint topic and perspective statistical model simultaneously uncovers two factors -- topical and ideologicalfactors -- that could account for why a word is used more frequently (emphasized) or less frequently (de-emphasized). A word can be emphasized because it represents what a text is about (topic), while a word can be also emphasized because authors holding a particular ideological perspective choose so.
The demo was shown in the Journalism 3G: The Future of Technology in the Field: A Symposium on Computation + Journalism, February 22-23, Georgia Institute of Technology, Atlanta, GA.