Jointly Learning Entity and Relation Representations for Entity Alignment

Abstract

Entity alignment is a viable means for integrating heterogeneous knowledge among different knowledge graphs (KGs). Recent developments in the field often take an embeddingbased approach to model the structural information of KGs so that entity alignment can be easily performed in the embedding space. However, most existing works do not explicitly utilize useful relation representations to assist in entity alignment, which, as we will show in the paper, is a simple yet effective way for improving entity alignment. This paper presents a novel joint learning framework for entity alignment. At the core of our approach is a Graph Convolutional Network (GCN) based framework for learning both entity and relation representations. Rather than relying on pre-aligned relation seeds to learn relation representations, we first approximate them using entity embeddings learned by the GCN. We then incorporate the relation approximation into entities to iteratively learn better representations for both. Experiments performed on three real-world cross-lingual datasets show that our approach substantially outperforms state-of-the-art entity alignment methods. Recent studies have shown that jointly modeling entities and relations in a single framework can arXiv:1909.09317v1 [cs.CL]

Relational Reflection Entity Alignment

“…Notice that our hypothesis is applicable to the alignment methods purely based on structural information (i.e., triples). Some methods [31,33] take entity names and pre-align them by machine translation or cross-lingual word embeddings. In these methods, GNNs play a role as noise smoothing rather than actual alignment.…”

“…• Semi-supervised: This kind of methods introduces semisupervision to generate extra structural data: Boot-EA [25], NAEA [35], TransEdge (semi), MRAEA (semi). • Textual: Besides the structural data, textual methods introduce entity names as additional input features: GMNN [33], RDGCN [30], HGCN [31], MRAEA (text) and DGMC [7].…”

Combining Graph-Based Learning With Automated Data Collection for Code Vulnerability Detection

“…Graph neural networks. GNNs have shown promising results in processing graph data structures for tasks like mining social networks [12], entity alignment [13], and binary similarity detection [14]. DEVIGN [15] is most closely related to our work.…”

“…It achieves this by leveraging the recently proposed gated graph neural networks (GGNNs) [11]. By directly operating on a graph representation, the graph neural networks (GNNs) have shown astounding successes in social networks [12], and knowledge graphs [13] and even compiled binaries [14]. While GNN provides a good starting point, applying it to develop a practical and efficient framework for software vulnerability detection is not trivial.…”

REA: Robust Cross-lingual Entity Alignment Between Knowledge Graphs

“…Then MuGNN [5] combined GNN with self attention to encode KGs, and contextual information also can be utilized with a GNN-based graph matching model [43]. Moreover, simple relation information can be encoded with the GNN models [40,41,45], and MRAEA [24] further takes the meta semantic information of relations into consideration.…”