Improving long-tail relation extraction via adaptive adjustment and causal inference

被引:4
作者
Tang, Jingyao [1 ]
Li, Lishuang [1 ]
Lu, Hongbin [1 ]
Zhang, Beibei [1 ]
Wu, Haiming [2 ]
机构
[1] Dalian Univ Technol, Sch Comp Sci & Technol, 2 Linggong Rd, Dalian 116024, Liaoning, Peoples R China
[2] Beijing Inst Technol, Sch Comp Sci & Technol, 5 South St, Beijing 100081, Peoples R China
基金
中国国家自然科学基金;
关键词
Long tail; Relation Extraction; Adaptive adjustment; Causal inference;
D O I
10.1016/j.neucom.2023.126563
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Extracting long-tail relations poses a significant challenge. Traditional models struggle with weak generalization on tail classes due to the limited sample size. To overcome the limitation, we propose a novel long-tail relation extraction model based on Adaptive Adjustment and Causal Inference (AACI). Specifically, AACI leverages class -adaptive adjustment terms to increase the relative margins between head and tail classes, improving the dis-criminability of tail classes and further enhancing their generalization. Moreover, the learning of our model may encounter multiple spurious correlations due to confounding variables. Therefore, we construct a Structural Causal Model (SCM) for AACI to formalize all spurious correlations and apply causal inference methods to eliminate negative effects of these correlations, thus improving the robustness of AACI. We evaluate our model on the NYT24 and NYT datasets. Our experiments demonstrate that AACI effectively modulates the class margins, eliminates the spurious correlations, and outperforms existing state-of-the-art methods.
引用
收藏
页数:12
相关论文
共 27 条
[1]   Learning Relation Prototype From Unlabeled Texts for Long-Tail Relation Extraction [J].
Cao, Yixin ;
Kuang, Jun ;
Gao, Ming ;
Zhou, Aoying ;
Wen, Yonggang ;
Chua, Tat-Seng .
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, 2023, 35 (02) :1761-1774
[2]   Causal Inference in Natural Language Processing: Estimation, Prediction, Interpretation and Beyond [J].
Feder, Amir ;
Keith, Katherine A. ;
Manzoor, Emaad ;
Pryzant, Reid ;
Sridhar, Dhanya ;
Wood-Doughty, Zach ;
Eisenstein, Jacob ;
Grimmer, Justin ;
Reichart, Roi ;
Roberts, Margaret E. ;
Stewart, Brandon M. ;
Veitch, Victor ;
Yang, Diyi .
TRANSACTIONS OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS, 2022, 10 :1138-1158
[3]  
Gardent Claire, 2017, 10 INT C NAT LANG GE, P124
[4]  
Glymour M., 2016, Causal Inference in Statistics: A Primer
[5]  
Han X, 2018, 2018 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (EMNLP 2018), P2236
[6]   Disentangling Label Distribution for Long-tailed Visual Recognition [J].
Hong, Youngkyu ;
Han, Seungju ;
Choi, Kwanghee ;
Seo, Seokjun ;
Kim, Beomsu ;
Chang, Buru .
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, :6622-6632
[7]  
Kang Bingyi, 2019, 8 INT C LEARN REPR I
[8]  
LeCun Y., 2006, A tutorial on energy-based learning. Predicting Structured Data
[9]  
Lei K, 2018, P 27 INT C COMP LING, P426
[10]  
Li Y., 2020, P 28 INT C COMP LING, P1653, DOI 10.18653/v1/2020.coling-main.145