Effective analysis of causes and inter-dependencies of parsing errors

Authors:
Tadayoshi Hara;Yusuke Miyao;Jun'ichi Tsujii
Affiliations:
University of Tokyo, Bunkyo-ku, Tokyo, Japan;University of Tokyo, Bunkyo-ku, Tokyo, Japan;University of Tokyo, Bunkyo-ku, Tokyo, Japan and University of Manchester and NaCTeM (National Center for Text Mining)
Venue:
IWPT '09 Proceedings of the 11th International Conference on Parsing Technologies
Year:
2009

Citing 5
Cited 1

The syntactic process

The syntactic process
The Penn Treebank: annotating predicate argument structure

HLT '94 Proceedings of the workshop on Human Language Technology
Probabilistic disambiguation models for wide-coverage HPSG parsing

ACL '05 Proceedings of the 43rd Annual Meeting on Association for Computational Linguistics
Extremely lexicalized models for accurate and fast HPSG parsing

EMNLP '06 Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing
Evaluating impact of re-training a lexical disambiguation model on domain adaptation of an HPSG parser

IWPT '07 Proceedings of the 10th International Conference on Parsing Technologies

Detecting dependency parse errors with minimal resources

IWPT '11 Proceedings of the 12th International Conference on Parsing Technologies

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose two methods for analyzing errors in parsing. One is to classify errors into categories which grammar developers can easily associate with defects in grammar or a parsing model and thus its improvement. The other is to discover inter-dependencies among errors, and thus grammar developers can focus on errors which are crucial for improving the performance of a parsing model. The first method uses patterns of errors to associate them with categories of causes for those errors, such as errors in scope determination of coordination, PP-attachment, identification of antecedent of relative clauses, etc. On the other hand, the second method, which is based on reparsing with one of observed errors corrected, assesses inter-dependencies among errors by examining which other errors were to be corrected as a result if a specific error was corrected. Experiments show that these two methods are complementary and by being combined, they can provide useful clues as to how to improve a given grammar.