Neighborhood Matching Network for Entity Alignment

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6477–6487 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 6477 Neighborhood Matching Network for Entity Alignment Yuting Wu1, Xiao Liu1, Yansong Feng1,2∗, Zheng Wang3 and Dongyan Zhao1,2 1Wangxuan Institute of Computer Technology, Peking University, China 2The MOE Key Laboratory of Computational Linguistics, Peking University, China 3School of Computing, University of Leeds, U.K. {wyting,lxlisa,fengyansong,zhaodongyan}@pku.edu.cn z.wang5@leeds.ac.uk Abstract Structural heterogeneity between knowledge graphs is an outstanding challenge for entity alignment. This paper presents Neighborhood Matching Network (NMN), a novel entity alignment framework for tackling the struc- tural heterogeneity challenge. NMN estimates the similarities between entities to capture both the topological structure and the neigh- borhood difference. It provides two innova- tive components for better learning representa- tions for entity alignment. It first uses a novel graph sampling method to distill a discrimi- native neighborhood for each entity. It then adopts a cross-graph neighborhood matching module to jointly encode the neighborhood dif- ference for a given entity pair. Such strategies allow NMN to effectively construct matching- oriented entity representations while ignoring noisy neighbors that have a negative impact on the alignment task. Extensive experiments performed on three entity alignment datasets show that NMN can well estimate the neigh- borhood similarity in more tough cases and significantly outperforms 12 previous state-of- the-art methods. 1 Introduction By aligning entities from different knowledge graphs (KGs) to the same real-world identity, entity alignment is a powerful technique for knowledge integration. Unfortunately, entity alignment is non- trivial because real-life KGs are often incomplete and different KGs typically have heterogeneous schemas. Consequently, equivalent entities from two KGs could have distinct surface forms or dis- similar neighborhood structures. In recent years, embedding-based methods have become the dominated approach for entity align- ment (Zhu et al., 2017; Pei et al., 2019a; Cao et al., 2019; Xu et al., 2019; Li et al., 2019a; Sun et al., ∗Corresponding author. 布鲁克林区 纽约 美国 子行政区 子行政区 纽约州 子行政区 最大城市 最大城市 Brooklyn Queens southwest New York City subdivision subdivision New York state Manhattan south Dan Houser residence northwest . . . birthPlace David Julius KG1: 布鲁克林区 # neighbors: 3 # common neighbors: 2 KG2: Brooklyn # neighbors: 21 # common neighbors: 2 Case 1: equivalent entities with different sizes of neighborhoods (a) 利物浦 英格兰 英国 子行政区 国家 利物浦大学 城市 Liverpool Labour Party (UK) executive England constituent country United Kingdom sovereign state . . . birthPlace George Harrison KG1: 利物浦 # neighbors: 20 # common neighbors: 3 KG2: Liverpool # neighbors: 22 # common neighbors: 3 Case 2: most common neighbors are not discriminative for alignment (b) 工党（英国） 行政 . . . Figure 1: Illustrative examples: two tough cases for en- tity alignment. Dashed rectangles denote the common neighbors between different KGs. 2020). Such approaches have the advantage of not relying on manually constructed features or rules (Mahdisoltani et al., 2015). Using a set of seed alignments, an embedding-based method mod- els the KG structures to automatically learn how to map the …

Content truncated to 500 words