Improving the Quality of Text Understanding by Delaying Ambiguity Resolution

Text Understanding systems often commit to a single best interpretation of a sentence before analyzing subsequent text. This interpretation is chosen by resolving ambiguous alternatives to the one with the highest confidence, given the context available at the time of commitment. Subsequent text, however, may contain information that changes the confidence of alternatives. This may especially be the case with multiple redundant texts on the same topic. Ideally, systems would delay choosing among ambiguous alternatives until more text has been read. One solution is to maintain multiple candidate interpretations of each sentence until the system acquires disambiguating evidence. Unfortunately, the number of alternatives explodes quickly. In this paper, we propose a packed graphical (PG) representation that can efficiently represent a large number of alternative interpretations along with dependencies among them. We also present an algorithm for combining multiple PG representations to h...

• Type: Journal
• Year: 2010
• Where: COLING
• Authors: Doo Soon Kim, Ken Barker, Bruce W. Porter
• Keywords: Alternatives, Ambiguous Alternatives, COLING 2010, Computational Linguistics, Subsequent Text