Scenario customization for information extraction
Scenario customization for information extraction
On building a more efficient grammar by exploiting types
Natural Language Engineering
Automatic pattern acquisition for Japanese information extraction
HLT '01 Proceedings of the first international conference on Human language technology research
An improved extraction pattern representation model for automatic IE pattern acquisition
ACL '03 Proceedings of the 41st Annual Meeting on Association for Computational Linguistics - Volume 1
The Stanford typed dependencies representation
CrossParser '08 Coling 2008: Proceedings of the workshop on Cross-Framework and Cross-Domain Parser Evaluation
Comparing information extraction pattern models
IEBeyondDoc '06 Proceedings of the Workshop on Information Extraction Beyond The Document
A comparative study of syntactic parsers for event extraction
BioNLP '10 Proceedings of the 2010 Workshop on Biomedical Natural Language Processing
Adapting a probabilistic disambiguation model of an HPSG parser to a new domain
IJCNLP'05 Proceedings of the Second international joint conference on Natural Language Processing
Analysis and improvement of minimally supervised machine learning for relation extraction
NLDB'09 Proceedings of the 14th international conference on Applications of Natural Language to Information Systems
Hi-index | 0.00 |
In this paper, we propose to use dependency graphs rather than trees as the interface between a parser and the rule acquisition module of a relation extraction (RE) system. Dependency graphs are much more expressive than trees and can easily be adapted to the output representations of various parsers, in particular those with richer semantics. Our approach is built on top of an existing minimally supervised machine learning system for relation extraction. We extend its original tree-based interface to a graph-based representation. In our experiments, we make use of two different dependency parsers and a deep HPSG parser. As expected, switching to a graph representation for the parsers outputting dependency trees does not have any impact on the RE results. But using the graph-based representation for the extraction with deep HPSG analyses improves both recall and f-score of the RE and enables the system to extract more relation instances of higher arity. Furthermore, we also compare the performance among these parsers with respect to their contribution to the RE task. In general, the robust dependency parsers are good in recall. However, the fine-grained deep syntactic parsing wins when it comes to precision.