High Recall Open IE for Relation Discovery

Hady Elsahar, Christophe Gravier, Frederique Laforest


Abstract
Relation Discovery discovers predicates (relation types) from a text corpus relying on the co-occurrence of two named entities in the same sentence. This is a very narrowing constraint: it represents only a small fraction of all relation mentions in practice. In this paper we propose a high recall approach for Open IE, which enables covering up to 16 times more sentences in a large corpus. Comparison against OpenIE systems shows that our proposed approach achieves 28% improvement over the highest recall OpenIE system and 6% improvement in precision than the same system.
Anthology ID:
I17-2039
Volume:
Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 2: Short Papers)
Month:
November
Year:
2017
Address:
Taipei, Taiwan
Editors:
Greg Kondrak, Taro Watanabe
Venue:
IJCNLP
SIG:
Publisher:
Asian Federation of Natural Language Processing
Note:
Pages:
228–233
Language:
URL:
https://aclanthology.org/I17-2039
DOI:
Bibkey:
Cite (ACL):
Hady Elsahar, Christophe Gravier, and Frederique Laforest. 2017. High Recall Open IE for Relation Discovery. In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 228–233, Taipei, Taiwan. Asian Federation of Natural Language Processing.
Cite (Informal):
High Recall Open IE for Relation Discovery (Elsahar et al., IJCNLP 2017)
Copy Citation:
PDF:
https://aclanthology.org/I17-2039.pdf
Data
New York Times Annotated Corpus