Adjacency List Oriented Relational Fact Extraction via Adaptive Multi-task Learning

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3075–3087 August 1–6, 2021. ©2021 Association for Computational Linguistics 3075 Adjacency List Oriented Relational Fact Extraction via Adaptive Multi-task Learning Fubang Zhao1∗, Zhuoren Jiang2∗†, Yangyang Kang1, Changlong Sun1, Xiaozhong Liu3 1Alibaba Group, Hangzhou, China 2School of Public Affairs, Zhejiang University, Hangzhou, China 3School of Informatics, Computing and Engineering, IUB, Bloomington, USA fubang.zfb@alibaba-inc.com, jiangzhuoren@zju.edu.cn yangyang.kangyy@alibaba-inc.com, changlong.scl@taobao.com liu237@indiana.edu Abstract Relational fact extraction aims to extract se- mantic triplets from unstructured text. In this work, we show that all of the relational fact extraction models can be organized accord- ing to a graph-oriented analytical perspective. An efficient model, aDjacency lIst oRiented rElational faCT (DIRECT), is proposed based on this analytical framework. To alleviate challenges of error propagation and sub-task loss equilibrium, DIRECT employs a novel adaptive multi-task learning strategy with dy- namic sub-task loss balancing. Extensive ex- periments are conducted on two benchmark datasets, and results prove that the proposed model outperforms a series of state-of-the-art (SoTA) models for relational triplet extraction. 1 Introduction Relational fact extraction, as an essential NLP task, is playing an increasingly important role in knowledge graph construction (Han et al., 2019; Distiawan et al., 2019). It aims to extract rela- tional triplet from the text. A relational triplet is in the form of (subject, relation, object) or (s, r, o) (Zeng et al., 2019). While various prior models proposed for relational fact extraction, few of them analyze this task from the perspective of output data structure. As shown in Figure 1, the relational fact extrac- tion can be characterized as a directed graph con- struction task, where graph representation flexibil- ity and heterogeneity accompany additional bene- faction. In practice, there are three common ways to represent graphs (Gross and Yellen, 2005): Edge List is utilized to predict a sequence of triplets (edges). The recent sequence-to-sequence based models, such as NovelTagging (Zheng et al., 2017), CopyRE (Zeng et al., 2018), CopyRL (Zeng ∗These two authors contributed equally to this research. †Zhuoren Jiang is the corresponding author Figure 1: Example of exploring the relational fact ex- traction task from the perspective of directed graph rep- resentation method as output data structure. et al., 2019), and PNDec (Nayak and Ng, 2020), fall into this category. Edge list is a simple and space-efficient way to represent a graph (Arifuzzaman and Khan, 2015). However, there are three problems. First, the triplet overlapping problem (Zeng et al., 2018). For instance, as shown in Figure 1, for triplets (Obama, nationality, USA) and (Obama, presi- dent of, USA), there are two types of relations be- tween the “Obama” and “USA”. If the model only generates one sequence from the text (Zheng et al., 2017), it may fail to identify the multi-relation be- tween entities. Second, to overcome the triplet over- lapping problem, the model may have to extract the triplet element repeatedly (Zeng et al., 2018), which will increase the extraction cost. Third, there could be an ordering problem (Zeng et al., 2019): for multiple triplets, …

Content truncated to 500 words