Paper: A Comparison Of Tagging Strategies For Statistical Information Extraction

ACL ID N06-2038
Title A Comparison Of Tagging Strategies For Statistical Information Extraction
Venue Human Language Technologies
Session Short Paper
Year 2006
Authors

There are several approaches that model information extraction as a to- ken classification task, using various tagging strategies to combine multiple tokens. We describe the tagging strate- gies that can be found in the litera- ture and evaluate their relative perfor- mances. We also introduce a new strat- egy, called Begin/After tagging or BIA, and show that it is competitive to the best other strategies.