Few-Shot Multi-Hop Relation Reasoning over Knowledge Bases

Chuxu Zhang; Lu Yu; Mandana Saebi; Meng Jiang; Nitesh Chawla

doi:10.18653/v1/2020.findings-emnlp.51

Few-Shot Multi-Hop Relation Reasoning over Knowledge Bases

Chuxu Zhang, Lu Yu, Mandana Saebi, Meng Jiang, Nitesh Chawla

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

18 Scopus citations

Abstract

Multi-hop relation reasoning over knowledge base is to generate effective and interpretable relation prediction through reasoning paths. The current methods usually require sufficient training data (fact triples) for each query relation, impairing their performances over few-shot relations (with limited triples) which are common in knowledge base. To this end, we propose FIRE, a novel few-shot multi-hop relation learning model. FIRE applies reinforcement learning to model the sequential steps of multi-hop reasoning, besides performs heterogeneous structure encoding and knowledge-aware search space pruning. The meta-learning technique is employed to optimize model parameters that could quickly adapt to few-shot relations. Empirical study on two datasets demonstrate that FIRE outperforms state-of-the-art methods

Original language	English (US)
Title of host publication	Findings of the Association for Computational Linguistics: EMNLP 2020
Publisher	Association for Computational Linguistics (ACL)
DOIs	https://doi.org/10.18653/v1/2020.findings-emnlp.51
State	Published - 2020

Access to Document

10.18653/v1/2020.findings-emnlp.51

Cite this

@inproceedings{983baae00e9c4ad09d32de1f77d3a65c,

title = "Few-Shot Multi-Hop Relation Reasoning over Knowledge Bases",

abstract = "Multi-hop relation reasoning over knowledge base is to generate effective and interpretable relation prediction through reasoning paths. The current methods usually require sufficient training data (fact triples) for each query relation, impairing their performances over few-shot relations (with limited triples) which are common in knowledge base. To this end, we propose FIRE, a novel few-shot multi-hop relation learning model. FIRE applies reinforcement learning to model the sequential steps of multi-hop reasoning, besides performs heterogeneous structure encoding and knowledge-aware search space pruning. The meta-learning technique is employed to optimize model parameters that could quickly adapt to few-shot relations. Empirical study on two datasets demonstrate that FIRE outperforms state-of-the-art methods",

author = "Chuxu Zhang and Lu Yu and Mandana Saebi and Meng Jiang and Nitesh Chawla",

note = "KAUST Repository Item: Exported on 2021-04-14 Acknowledgements: This work was supported in part by National Science Foundation grants CCI-1925607 and IIS1849816. We also thank the anonymous reviewers for their valuable comments and helpful suggestions.",

year = "2020",

doi = "10.18653/v1/2020.findings-emnlp.51",

language = "English (US)",

booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2020",

publisher = "Association for Computational Linguistics (ACL)",

address = "United States",

}

TY - GEN

T1 - Few-Shot Multi-Hop Relation Reasoning over Knowledge Bases

AU - Zhang, Chuxu

AU - Yu, Lu

AU - Saebi, Mandana

AU - Jiang, Meng

AU - Chawla, Nitesh

N1 - KAUST Repository Item: Exported on 2021-04-14 Acknowledgements: This work was supported in part by National Science Foundation grants CCI-1925607 and IIS1849816. We also thank the anonymous reviewers for their valuable comments and helpful suggestions.

PY - 2020

Y1 - 2020

N2 - Multi-hop relation reasoning over knowledge base is to generate effective and interpretable relation prediction through reasoning paths. The current methods usually require sufficient training data (fact triples) for each query relation, impairing their performances over few-shot relations (with limited triples) which are common in knowledge base. To this end, we propose FIRE, a novel few-shot multi-hop relation learning model. FIRE applies reinforcement learning to model the sequential steps of multi-hop reasoning, besides performs heterogeneous structure encoding and knowledge-aware search space pruning. The meta-learning technique is employed to optimize model parameters that could quickly adapt to few-shot relations. Empirical study on two datasets demonstrate that FIRE outperforms state-of-the-art methods

AB - Multi-hop relation reasoning over knowledge base is to generate effective and interpretable relation prediction through reasoning paths. The current methods usually require sufficient training data (fact triples) for each query relation, impairing their performances over few-shot relations (with limited triples) which are common in knowledge base. To this end, we propose FIRE, a novel few-shot multi-hop relation learning model. FIRE applies reinforcement learning to model the sequential steps of multi-hop reasoning, besides performs heterogeneous structure encoding and knowledge-aware search space pruning. The meta-learning technique is employed to optimize model parameters that could quickly adapt to few-shot relations. Empirical study on two datasets demonstrate that FIRE outperforms state-of-the-art methods

UR - http://hdl.handle.net/10754/667823

UR - https://www.aclweb.org/anthology/2020.findings-emnlp.51

U2 - 10.18653/v1/2020.findings-emnlp.51

DO - 10.18653/v1/2020.findings-emnlp.51

M3 - Conference contribution

BT - Findings of the Association for Computational Linguistics: EMNLP 2020

PB - Association for Computational Linguistics (ACL)

ER -

Few-Shot Multi-Hop Relation Reasoning over Knowledge Bases

Abstract

Access to Document

Other files and links

Fingerprint

Cite this