English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent work on knowledge graph completion (KGC) focused on learningembeddings of entities and relations in knowledge graphs. These embeddingmethods require that all test entities are observed at training time, resultingin a time-consuming retraining process for out-of-knowledge-graph (OOKG)entities. To address this issue, current inductive knowledge embedding methodsemploy graph neural networks (GNNs) to represent unseen entities by aggregatinginformation of known neighbors. They face three important challenges: (i) datasparsity, (ii) the presence of complex patterns in knowledge graphs (e.g.,inter-rule correlations), and (iii) the presence of interactions among rulemining, rule inference, and embedding. In this paper, we propose a virtualneighbor network with inter-rule correlations (VNC) that consists of threestages: (i) rule mining, (ii) rule inference, and (iii) embedding. In the rulemining process, to identify complex patterns in knowledge graphs, both logicrules and inter-rule correlations are extracted from knowledge graphs based onoperations over relation embeddings. To reduce data sparsity, virtual neighborsfor OOKG entities are predicted and assigned soft labels by optimizing arule-constrained problem. We also devise an iterative framework to capture theunderlying relations between rule learning and embedding learning. In ourexperiments, results on both link prediction and triple classification tasksshow that the proposed VNC framework achieves state-of-the-art performance onfour widely-used knowledge graphs. Further analysis reveals that VNC is robustto the proportion of unseen entities and effectively mitigates data sparsity.

Iteratively Learning Representations for Unseen Entities with Inter-Rule Correlations