Relation extraction is often challenged by insufficient labeled data. Previous methods exploit knowledge from unlabeled data by generating pseudo labels in a self-training pipeline, which suffers a gradual drift problem. Logic rules, a transferable and explainable form of expert knowledge, have achieved promising success by improving the model with weak labels. But manually writing comprehensive rules set is challenging and tedious. To alleviate the human labor of writing high-quality rules, in this work, we propose ARIA, an Automatic task-specific Rules distilling framework. Specifically, we guide the pre-trained language model to reason rules as experts and compose them into robust compound rules for data labeling. Besides, ARIA could continuously enrich the rules set to power the labeling ability by discovering reliable model-labeled data for distinguishable rules generation. Experiments on two public datasets demonstrate the effectiveness of ARIA in a low-resource scenario.
%0 Conference Paper
%1 lu2023reasoning
%A Lu, Yilin
%A Li, Juncheng
%A Wang, Xiaoqiang
%A Shi, Haochen
%A Chen, Tao
%A Tang, Siliang
%B Findings of the Association for Computational Linguistics: EMNLP 2023
%C Singapore
%D 2023
%E Bouamor, Houda
%E Pino, Juan
%E Bali, Kalika
%I Association for Computational Linguistics
%K low-resource nlp relationclassification relations
%P 7447--7457
%R 10.18653/v1/2023.findings-emnlp.499
%T Reasoning Makes Good Annotators : An Automatic Task-specific Rules Distilling Framework for Low-resource Relation Extraction
%U https://aclanthology.org/2023.findings-emnlp.499
%X Relation extraction is often challenged by insufficient labeled data. Previous methods exploit knowledge from unlabeled data by generating pseudo labels in a self-training pipeline, which suffers a gradual drift problem. Logic rules, a transferable and explainable form of expert knowledge, have achieved promising success by improving the model with weak labels. But manually writing comprehensive rules set is challenging and tedious. To alleviate the human labor of writing high-quality rules, in this work, we propose ARIA, an Automatic task-specific Rules distilling framework. Specifically, we guide the pre-trained language model to reason rules as experts and compose them into robust compound rules for data labeling. Besides, ARIA could continuously enrich the rules set to power the labeling ability by discovering reliable model-labeled data for distinguishable rules generation. Experiments on two public datasets demonstrate the effectiveness of ARIA in a low-resource scenario.
@inproceedings{lu2023reasoning,
abstract = {Relation extraction is often challenged by insufficient labeled data. Previous methods exploit knowledge from unlabeled data by generating pseudo labels in a self-training pipeline, which suffers a gradual drift problem. Logic rules, a transferable and explainable form of expert knowledge, have achieved promising success by improving the model with weak labels. But manually writing comprehensive rules set is challenging and tedious. To alleviate the human labor of writing high-quality rules, in this work, we propose ARIA, an Automatic task-specific Rules distilling framework. Specifically, we guide the pre-trained language model to reason rules as experts and compose them into robust compound rules for data labeling. Besides, ARIA could continuously enrich the rules set to power the labeling ability by discovering reliable model-labeled data for distinguishable rules generation. Experiments on two public datasets demonstrate the effectiveness of ARIA in a low-resource scenario.},
added-at = {2024-01-09T13:21:11.000+0100},
address = {Singapore},
author = {Lu, Yilin and Li, Juncheng and Wang, Xiaoqiang and Shi, Haochen and Chen, Tao and Tang, Siliang},
biburl = {https://www.bibsonomy.org/bibtex/2ffef7b90340e5c2a956e16b5de5112be/albinzehe},
booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2023},
doi = {10.18653/v1/2023.findings-emnlp.499},
editor = {Bouamor, Houda and Pino, Juan and Bali, Kalika},
interhash = {5dac71f12b4bdd24c8a9b3285e4d3e69},
intrahash = {ffef7b90340e5c2a956e16b5de5112be},
keywords = {low-resource nlp relationclassification relations},
month = dec,
pages = {7447--7457},
publisher = {Association for Computational Linguistics},
timestamp = {2024-01-09T13:21:11.000+0100},
title = {Reasoning Makes Good Annotators : An Automatic Task-specific Rules Distilling Framework for Low-resource Relation Extraction},
url = {https://aclanthology.org/2023.findings-emnlp.499},
year = 2023
}