Research in information extraction
Ralph Grishman · 1996
Information extraction involves picking out specified types of information from natural language text. Recent Message Understanding Conferences [1,2,3] have developed a spectrum of such tasks, and we have worked on two of them, at opposite ends of the spectrum: the named entity task, which involves identifying and classifying names, and the scenario template task, which involves extracting critical information (participants, location, date, etc.) about specified classes of events.