ICSI-CRF: the generation of references to the main subject and named entities using conditional random fields

Authors:
Benoit Favre;Bernd Bohnet
Affiliations:
International Computer Science Institute, Berkeley, CA;International Computer Science Institute, Berkeley, CA
Venue:
UCNLG+Sum '09 Proceedings of the 2009 Workshop on Language Generation and Summarisation
Year:
2009

Citing 2
Cited 2

Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data

ICML '01 Proceedings of the Eighteenth International Conference on Machine Learning
Generation of repeated references to discourse entities

ENLG '07 Proceedings of the Eleventh European Workshop on Natural Language Generation

The UMUS system for named entity generation at GREC 2010

INLG '10 Proceedings of the 6th International Natural Language Generation Conference
UDel: refining a method of named entity generation

INLG '10 Proceedings of the 6th International Natural Language Generation Conference

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we describe our contribution to the Generation Challenge 2009 for the tasks of generating Referring Expressions to the Main Subject References (MSR) and Named Entities Generation (NEG). To generate the referring expressions, we employ the Conditional Random Fields (CRF) learning technique due to the fact that the selection of an expression depends on the selection of the previous references. CRFs fit very well to this task since they are designed for the labeling of sequences. For the MSR task, our system has a String Accuracy of 0.68 and a REG08-Type Accuracy of 0.76 and for the NEG task a String Accuracy of 0.79 and REG08-Type Accuracy of 0.83.