Minimally supervised novel relation extraction using a latent relational mapping

Danushka Bollegala, Yutaka Matsuo, Mitsuru Ishizuka

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

The World Wide Web includes semantic relations of numerous types that exist among different entities. Extracting the relations that exist between two entities is an important step in various Web-related tasks such as information retrieval (IR), information extraction, and social network extraction. A supervised relation extraction system that is trained to extract a particular relation type (source relation) might not accurately extract a new type of a relation (target 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. First, to represent a semantic relation that exists between two entities, we extract lexical and syntactic patterns from contexts in which those two entities co-occur. Then, we construct a bipartite graph between relation-specific (RS) and relation-independent (RI) patterns. Spectral clustering is performed on the bipartite graph to compute a lower dimensional projection. Second, we train a classifier for the target relation type using a small number of labeled instances. To account for the lack of target relation training instances, we present a one-sided under sampling method. We evaluate the proposed method using a data set that contains 2,000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macroaverage F-score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly supervised relation extraction method.

Original languageEnglish
Article number6095557
Pages (from-to)419-432
Number of pages14
JournalIEEE Transactions on Knowledge and Data Engineering
Volume25
Issue number2
DOIs
Publication statusPublished - 2013
Externally publishedYes

Fingerprint

Classifiers
Semantics
Sampling
Syntactics
Information retrieval
World Wide Web

Keywords

  • domain adaptation
  • Relation extraction
  • Web mining

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Information Systems
  • Computer Science Applications

Cite this

Minimally supervised novel relation extraction using a latent relational mapping. / Bollegala, Danushka; Matsuo, Yutaka; Ishizuka, Mitsuru.

In: IEEE Transactions on Knowledge and Data Engineering, Vol. 25, No. 2, 6095557, 2013, p. 419-432.

Research output: Contribution to journalArticle

Bollegala, Danushka ; Matsuo, Yutaka ; Ishizuka, Mitsuru. / Minimally supervised novel relation extraction using a latent relational mapping. In: IEEE Transactions on Knowledge and Data Engineering. 2013 ; Vol. 25, No. 2. pp. 419-432.
@article{1cb7675f56234ec286620a035f3f7d4b,
title = "Minimally supervised novel relation extraction using a latent relational mapping",
abstract = "The World Wide Web includes semantic relations of numerous types that exist among different entities. Extracting the relations that exist between two entities is an important step in various Web-related tasks such as information retrieval (IR), information extraction, and social network extraction. A supervised relation extraction system that is trained to extract a particular relation type (source relation) might not accurately extract a new type of a relation (target 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. First, to represent a semantic relation that exists between two entities, we extract lexical and syntactic patterns from contexts in which those two entities co-occur. Then, we construct a bipartite graph between relation-specific (RS) and relation-independent (RI) patterns. Spectral clustering is performed on the bipartite graph to compute a lower dimensional projection. Second, we train a classifier for the target relation type using a small number of labeled instances. To account for the lack of target relation training instances, we present a one-sided under sampling method. We evaluate the proposed method using a data set that contains 2,000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macroaverage F-score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly supervised relation extraction method.",
keywords = "domain adaptation, Relation extraction, Web mining",
author = "Danushka Bollegala and Yutaka Matsuo and Mitsuru Ishizuka",
year = "2013",
doi = "10.1109/TKDE.2011.250",
language = "English",
volume = "25",
pages = "419--432",
journal = "IEEE Transactions on Knowledge and Data Engineering",
issn = "1041-4347",
publisher = "IEEE Computer Society",
number = "2",

}

TY - JOUR

T1 - Minimally supervised novel relation extraction using a latent relational mapping

AU - Bollegala, Danushka

AU - Matsuo, Yutaka

AU - Ishizuka, Mitsuru

PY - 2013

Y1 - 2013

N2 - The World Wide Web includes semantic relations of numerous types that exist among different entities. Extracting the relations that exist between two entities is an important step in various Web-related tasks such as information retrieval (IR), information extraction, and social network extraction. A supervised relation extraction system that is trained to extract a particular relation type (source relation) might not accurately extract a new type of a relation (target 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. First, to represent a semantic relation that exists between two entities, we extract lexical and syntactic patterns from contexts in which those two entities co-occur. Then, we construct a bipartite graph between relation-specific (RS) and relation-independent (RI) patterns. Spectral clustering is performed on the bipartite graph to compute a lower dimensional projection. Second, we train a classifier for the target relation type using a small number of labeled instances. To account for the lack of target relation training instances, we present a one-sided under sampling method. We evaluate the proposed method using a data set that contains 2,000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macroaverage F-score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly supervised relation extraction method.

AB - The World Wide Web includes semantic relations of numerous types that exist among different entities. Extracting the relations that exist between two entities is an important step in various Web-related tasks such as information retrieval (IR), information extraction, and social network extraction. A supervised relation extraction system that is trained to extract a particular relation type (source relation) might not accurately extract a new type of a relation (target 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. First, to represent a semantic relation that exists between two entities, we extract lexical and syntactic patterns from contexts in which those two entities co-occur. Then, we construct a bipartite graph between relation-specific (RS) and relation-independent (RI) patterns. Spectral clustering is performed on the bipartite graph to compute a lower dimensional projection. Second, we train a classifier for the target relation type using a small number of labeled instances. To account for the lack of target relation training instances, we present a one-sided under sampling method. We evaluate the proposed method using a data set that contains 2,000 instances for 20 different relation types. Our experimental results show that the proposed method achieves a statistically significant macroaverage F-score of 62.77. Moreover, the proposed method outperforms numerous baselines and a previously proposed weakly supervised relation extraction method.

KW - domain adaptation

KW - Relation extraction

KW - Web mining

UR - http://www.scopus.com/inward/record.url?scp=84871658320&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84871658320&partnerID=8YFLogxK

U2 - 10.1109/TKDE.2011.250

DO - 10.1109/TKDE.2011.250

M3 - Article

AN - SCOPUS:84871658320

VL - 25

SP - 419

EP - 432

JO - IEEE Transactions on Knowledge and Data Engineering

JF - IEEE Transactions on Knowledge and Data Engineering

SN - 1041-4347

IS - 2

M1 - 6095557

ER -