Iteratively Learning Representations for Unseen Entities with Inter-Rule Correlations

Abstract

Recent work on knowledge graph completion (KGC) focuses on acquiring embeddings of entities and relations in knowledge graphs. These embedding methods necessitate that all test entities be present during the training phase, resulting in a time-consuming retraining process for out-of-knowledge-graph (OOKG) entities. To tackle this predicament, current inductive methods employ graph neural networks (GNNs) to represent unseen entities by aggregating information of the known neighbors, and enhance the performance with additional information, such as attention mechanisms or logic rules. Nonetheless, Two key challenges continue to persist: (i) identifying inter-rule correlations to further facilitate the inference process, and (ii) capturing interactions among rule mining, rule inference, and embedding to enhance both rule and embedding learning. In this paper, we propose a virtual neighbor network with inter-rule correlations (VNC) to address the above challenges. VNC consists of three main components: (i) rule mining, (ii) rule inference, and (iii) embedding. To identify useful complex patterns in knowledge graphs, both logic rules and inter-rule correlations are extracted from knowledge graphs based on operations over relation embeddings. To reduce data sparsity, virtual networks for OOKG entities are predicted and assigned soft labels by optimizing a rule-constrained problem. We also devise an iterative framework to capture the underlying interactions between rule and embedding learning. Experimental results on both link prediction and triple classification tasks show that the proposed VNC framework achieves state-of-the-art performance on four widely-used knowledge graphs. Our code and data are available at https://github.com/WZH-NLP/OOKG.