Diversity of Scenarios in Information Extraction

Silja Huttunen (Courant Institute of Mathematical Sciences New York University)

Roman Yangarber (Courant Institute of Mathematical Sciences New York University)

Ralph Grishman (Courant Institute of Mathematical Sciences New York University)

WO16: Applications Based On Written LRs

This paper discusses/presents problems of template structure for Information Extraction. We investigate these problems in the context of two new Information Extraction scenarios which are linguistically and structurally more challenging than the traditional MUC scenarios. By a scenario we mean a predefined set of facts to be extracted from text. Traditional views on event structure and template design are not adequate for the more complex scenarios. We identify two structural factors that contribute to the complexity of a scenario: first, the scattering of events in text, and second, inclusion relationship between events. These factors cause difficulty in representing the facts in an unambiguous way. Traditional views on event structure and template design are not adequate for the more complex scenarios. We propose that these kinds of event relationships can be better described with a modular, hierarchical model.

Information extraction, Scenario, Complexity, Text type

335.pdf