Relation adaptation: Learning to extract novel relations with minimum supervision

Danushka Bollegala, Yutaka Matsuo, Mitsuru Ishizuka

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

Extracting the relations that exist between two entities is an important step in numerous Web-related tasks such as information extraction. A supervised relation extraction system that is trained to extract a particular relation type might not accurately extract a new type of a relation for which it has not been trained. However, it is costly to create training data manually for every new relation type that one might want to extract. We propose a method to adapt an existing relation extraction system to extract new relation types with minimum supervision. Our proposed method comprises two stages: learning a lower-dimensional projection between different relations, and learning a relational classifier for the target relation type with instance sampling. We evaluate the proposed method using a dataset that contains 2000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macro-average F -score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly-supervised relation extraction method.

Original languageEnglish
Title of host publicationIJCAI International Joint Conference on Artificial Intelligence
Pages2205-2210
Number of pages6
DOIs
Publication statusPublished - 2011
Externally publishedYes
Event22nd International Joint Conference on Artificial Intelligence, IJCAI 2011 - Barcelona, Catalonia
Duration: 2011 Jul 162011 Jul 22

Other

Other22nd International Joint Conference on Artificial Intelligence, IJCAI 2011
CityBarcelona, Catalonia
Period11/7/1611/7/22

Fingerprint

Macros
Classifiers
Sampling

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Bollegala, D., Matsuo, Y., & Ishizuka, M. (2011). Relation adaptation: Learning to extract novel relations with minimum supervision. In IJCAI International Joint Conference on Artificial Intelligence (pp. 2205-2210) https://doi.org/10.5591/978-1-57735-516-8/IJCAI11-368

Relation adaptation : Learning to extract novel relations with minimum supervision. / Bollegala, Danushka; Matsuo, Yutaka; Ishizuka, Mitsuru.

IJCAI International Joint Conference on Artificial Intelligence. 2011. p. 2205-2210.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Bollegala, D, Matsuo, Y & Ishizuka, M 2011, Relation adaptation: Learning to extract novel relations with minimum supervision. in IJCAI International Joint Conference on Artificial Intelligence. pp. 2205-2210, 22nd International Joint Conference on Artificial Intelligence, IJCAI 2011, Barcelona, Catalonia, 11/7/16. https://doi.org/10.5591/978-1-57735-516-8/IJCAI11-368
Bollegala D, Matsuo Y, Ishizuka M. Relation adaptation: Learning to extract novel relations with minimum supervision. In IJCAI International Joint Conference on Artificial Intelligence. 2011. p. 2205-2210 https://doi.org/10.5591/978-1-57735-516-8/IJCAI11-368
Bollegala, Danushka ; Matsuo, Yutaka ; Ishizuka, Mitsuru. / Relation adaptation : Learning to extract novel relations with minimum supervision. IJCAI International Joint Conference on Artificial Intelligence. 2011. pp. 2205-2210
@inproceedings{40fc3e3a3f1f40ae88a0fab133bc2499,
title = "Relation adaptation: Learning to extract novel relations with minimum supervision",
abstract = "Extracting the relations that exist between two entities is an important step in numerous Web-related tasks such as information extraction. A supervised relation extraction system that is trained to extract a particular relation type might not accurately extract a new type of a relation for which it has not been trained. However, it is costly to create training data manually for every new relation type that one might want to extract. We propose a method to adapt an existing relation extraction system to extract new relation types with minimum supervision. Our proposed method comprises two stages: learning a lower-dimensional projection between different relations, and learning a relational classifier for the target relation type with instance sampling. We evaluate the proposed method using a dataset that contains 2000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macro-average F -score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly-supervised relation extraction method.",
author = "Danushka Bollegala and Yutaka Matsuo and Mitsuru Ishizuka",
year = "2011",
doi = "10.5591/978-1-57735-516-8/IJCAI11-368",
language = "English",
isbn = "9781577355120",
pages = "2205--2210",
booktitle = "IJCAI International Joint Conference on Artificial Intelligence",

}

TY - GEN

T1 - Relation adaptation

T2 - Learning to extract novel relations with minimum supervision

AU - Bollegala, Danushka

AU - Matsuo, Yutaka

AU - Ishizuka, Mitsuru

PY - 2011

Y1 - 2011

N2 - Extracting the relations that exist between two entities is an important step in numerous Web-related tasks such as information extraction. A supervised relation extraction system that is trained to extract a particular relation type might not accurately extract a new type of a relation for which it has not been trained. However, it is costly to create training data manually for every new relation type that one might want to extract. We propose a method to adapt an existing relation extraction system to extract new relation types with minimum supervision. Our proposed method comprises two stages: learning a lower-dimensional projection between different relations, and learning a relational classifier for the target relation type with instance sampling. We evaluate the proposed method using a dataset that contains 2000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macro-average F -score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly-supervised relation extraction method.

AB - Extracting the relations that exist between two entities is an important step in numerous Web-related tasks such as information extraction. A supervised relation extraction system that is trained to extract a particular relation type might not accurately extract a new type of a relation for which it has not been trained. However, it is costly to create training data manually for every new relation type that one might want to extract. We propose a method to adapt an existing relation extraction system to extract new relation types with minimum supervision. Our proposed method comprises two stages: learning a lower-dimensional projection between different relations, and learning a relational classifier for the target relation type with instance sampling. We evaluate the proposed method using a dataset that contains 2000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macro-average F -score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly-supervised relation extraction method.

UR - http://www.scopus.com/inward/record.url?scp=84881049139&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84881049139&partnerID=8YFLogxK

U2 - 10.5591/978-1-57735-516-8/IJCAI11-368

DO - 10.5591/978-1-57735-516-8/IJCAI11-368

M3 - Conference contribution

AN - SCOPUS:84881049139

SN - 9781577355120

SP - 2205

EP - 2210

BT - IJCAI International Joint Conference on Artificial Intelligence

ER -