FactBrowser Demonstration

The FactBrowser demonstration illustrates automatic database update from live feeds based on information extraction from text and the ability to browse the resulting database for unexpected connections. The technology used has four interesting features:1. The demonstration employs a light architecture based on the Web; using an XML-based client-server architecture, the graphical user interface requires only Internet Explorer 5.0 or higher. No application code resides on the client.2. A permanent database grows based on cross-document entity tracking and accumulating facts.3. The database is updated daily based on automatic processing of documents distributed by the Foreign Broadcasting Information Service (FBIS). Document capture and database update are fully automatic, requiring no human intervention.4. The following key components: name finding, parsing, and pronoun resolution are all based on the trained, language-independent statistical modeling techniques.The strategic focus throughout the design of FactBrowser has been on producing high precision output so as to maintain quality in the data base.