Multilingual Knowledge Graph Completion via Ensemble Knowledge Transfer

Findings of the Association for Computational Linguistics: EMNLP 2020, pages 3227–3238 November 16 - 20, 2020. c⃝2020 Association for Computational Linguistics 3227 Multilingual Knowledge Graph Completion via Ensemble Knowledge Transfer Xuelu Chen1, Muhao Chen1,2,3, Changjun Fan4∗, Ankith Uppunda1, Yizhou Sun1 & Carlo Zaniolo1 1Department of Computer Science, UCLA, USA 2Information Sciences Institute, USC, USA 3Department of Computer and Information Science, UPenn, USA 4College of Systems Engineering, NUDT, China shirleychen@ucla.edu; muhaoche@usc.edu; fanchangjun@nudt.edu.cn; {auppunda, yzsun, zaniolo}@cs.ucla.edu Abstract Predicting missing facts in a knowledge graph (KG) is a crucial task in knowledge base con- struction and reasoning, and it has been the subject of much research in recent works us- ing KG embeddings. While existing KG em- bedding approaches mainly learn and predict facts within a single KG, a more plausible so- lution would benefit from the knowledge in multiple language-specific KGs, considering that different KGs have their own strengths and limitations on data quality and coverage. This is quite challenging, since the transfer of knowledge among multiple independently maintained KGs is often hindered by the in- sufficiency of alignment information and the inconsistency of described facts. In this pa- per, we propose KEnS, a novel framework for embedding learning and ensemble knowl- edge transfer across a number of language- specific KGs. KEnS embeds all KGs in a shared embedding space, where the associ- ation of entities is captured based on self- learning. Then, KEnS performs ensemble in- ference to combine prediction results from em- beddings of multiple language-specific KGs, for which multiple ensemble techniques are investigated. Experiments on five real-world language-specific KGs show that KEnS con- sistently improves state-of-the-art methods on KG completion, via effectively identifying and leveraging complementary knowledge. 1 Introduction Knowledge graphs (KGs) store structured repre- sentations of real-world entities and relations, con- stituting actionable knowledge that is crucial to various knowledge-driven applications (Koncel- Kedziorski et al., 2019; Chen et al., 2018a; Bordes et al., 2014). Recently, extensive efforts have been invested in KG embedding models, which encode ∗ This work was done when this author was visiting University of California, Los Angeles. entities as low-dimensional vectors and capture re- lations as algebraic operations on entity vectors. These models provide a beneficial tool to complete KGs by discovering previously unknown knowl- edge from latent representations of observed facts. Representative models including translational mod- els (Bordes et al., 2013; Wang et al., 2014) and bilinear models (Yang et al., 2015; Trouillon et al., 2016) have achieved satisfactory performance in predicting missing facts. Existing methods mainly investigate KG comple- tion within a single monolingual KG. As different language-specific KGs have their own strengths and limitations on data quality and coverage, we investigate a more natural solution, which seeks to combine embedding models of multiple KGs in an ensemble-like manner. This approach offers several potential benefits. First, embedding models of well- populated KGs (e.g. English KGs) are expected to capture richer knowledge because of better data quality and denser graph structures. Therefore, they would provide ampler signals to facilitate inferring missing facts on …

Content truncated to 500 words