Text Extraction

also called Information Extraction

• The MUCs did Text Extraction.
• Roughly, a particular database template needs to be filled.
• A system is designed to read (newspaper) articles and when appropriate, fill in these templates.
• Example MUC domains were Central American Terrorist events, rocket launch events, and movements between companies of corporate executives.
• These systems aren't perfect, but perform quite well.
• Moreover, they are really fast.
• Note that they are domain specific and a new system needs to built for a new domain.
• Performance is better in simpler domains.