Interpretable and Low-Resource Entity Matching via Decoupling Feature Learning from Decision Making

Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 2770–2781 August 1–6, 2021. ©2021 Association for Computational Linguistics 2770 Interpretable and Low-Resource Entity Matching via Decoupling Feature Learning from Decision Making Zijun Yao1,2 Chengjiang Li1,2 Tiansi Dong3 Xin Lv1,2 Jifan Yu1,2 Lei Hou1,2∗ Juanzi Li1,2 Yichi Zhang4 Zelin Dai4 1Department of Computer Science and Technology, BNRist; 2KIRC, Institute for Artificial Intelligence Tsinghua University, Beijing 100084, China 3B-IT, University of Bonn, Germany 4Alibaba Group, Hangzhou, China {yaozj20@mails., houlei@}tsinghua.edu.cn dongt@bit.uni-bonn.de Abstract Entity Matching (EM) aims at recognizing en- tity records that denote the same real-world ob- ject. Neural EM models learn vector represen- tation of entity descriptions and match entities end-to-end. Though robust, these methods re- quire many annotated resources for training, and lack of interpretability. In this paper, we propose a novel EM framework that consists of Heterogeneous Information Fusion (HIF) and Key Attribute Tree (KAT) Induction to decou- ple feature representation from matching deci- sion. Using self-supervised learning and mask mechanism in pre-trained language modeling, HIF learns the embeddings of noisy attribute values by inter-attribute attention with unla- beled data. Using a set of comparison fea- tures and a limited amount of annotated data, KAT Induction learns an efficient decision tree that can be interpreted by generating entity matching rules whose structure is advocated by domain experts. Experiments on 6 pub- lic datasets and 3 industrial datasets show that our method is highly efficient and outperforms SOTA EM models in most cases. Our codes and datasets can be obtained from https:// github.com/THU-KEG/HIF-KAT. 1 Introduction Entity Matching (EM) aims at identifying whether two records from different sources refer to the same real-world entity. This is a fundamental research task in knowledge graph integration (Dong et al., 2014; Daniel et al., 2020; Christophides et al., 2015; Christen, 2012) and text mining (Zhao et al., 2014). In real applications, it is not easy to decide whether two records with ad hoc linguistic descriptions refer to the same entity. In Figure 1, e2 and e3 refer to the same publication, while e1 refers to a different ∗ Corresponding to L.Hou (houlei@tsinghua.edu.cn) Title Author Venue Conference (redundant) 𝑒! Data Mining Techniques missing SIGMOD Conference International Conference on Management of Data 𝑒" Data Mining: Concepts and Techniques J. Han, J. Pei, M. Kamber SIGMOD Record missing 𝑒# Data mining: Concepts & Techniques by Jiawei Han misplaced ACM SIGMOD Record missing Figure 1: Published papers as entity records. one. Venues of e2 and e3 have different expressions; Authors of e3 is misplaced in its Title field. Early works include feature engineering (Wang et al., 2011) and rule matching (Singh et al., 2017; Fan et al., 2009). Recently, the robustness of En- tity Matching has been improved by deep learning models, such as distributed representation based models (Ebraheem et al., 2018), attention based models (Mudgal et al., 2018; Fu et al., 2019, 2020), and pre-trained language model based models (Li et al., 2020). Nevertheless, …

Content truncated to 500 words