Many machine learning methods have recently been applied to natural language processing tasks. Among them, the Winnow algorithm has been argued to be particularly suitable for NLP problems, due to its robustness to irrelevant features. However in theory, Winnow may not converge for non-separable data. To remedy this problem, a modification called regularized Winnow has been proposed. In this paper, we apply this new method to text chunking. We show that this method achieves state of the art performance with significantly less computation than previous approaches.
Text Chunking using Regularized Winnow
Tong Zhang,Fred J. Damerau,David E. Johnson
Published 2001 in Annual Meeting of the Association for Computational Linguistics
ABSTRACT
PUBLICATION RECORD
- Publication year
2001
- Venue
Annual Meeting of the Association for Computational Linguistics
- Publication date
2001-07-06
- Fields of study
Computer Science
- Identifiers
- External record
- Source metadata
Semantic Scholar
CITATION MAP
EXTRACTION MAP
CLAIMS
CONCEPTS
- computation
The amount of processing required to train or apply a method on the text chunking task.
Aliases: computational cost
- non-separable data
Training data that cannot be perfectly separated by a linear classifier.
- previous approaches
Earlier text chunking methods used as comparison baselines in the paper.
Aliases: prior approaches, earlier approaches
- regularized winnow
A modified Winnow variant introduced to handle non-separable data more robustly.
- state-of-the-art performance
The highest reported level of effectiveness on the evaluated text chunking task.
Aliases: SOTA performance, SOTA
- text chunking
The natural language processing task of segmenting text into syntactic chunks.
Aliases: chunking
- winnow algorithm
A linear learning algorithm used for feature-based classification and discussed here as the method being regularized.
Aliases: Winnow
REFERENCES
Showing 1-13 of 13 references · Page 1 of 1
CITED BY
Showing 1-49 of 49 citing papers · Page 1 of 1