Trans4E: Link prediction on scholarly knowledge graphs

The incompleteness of Knowledge Graphs (KGs) is a crucial issue affecting the quality AI-based services. In scholarly domain, KGs describing research publications typically lack important information, hindering our ability to analyse and predict dynamics. recent years, link prediction approaches based on Graph Embedding models became first aid for this issue. work, we present Trans4E, novel embedding model that particularly fit which include N M relations with N$\gg$M. This typical categorize large number entities (e.g., articles, patents, persons) according relatively small set categories. Trans4E was applied two large-scale knowledge graphs, Academia/Industry DynAmics (AIDA) Microsoft Academic (MAG), completing information about Fields Study 'neural networks', 'machine learning', 'artificial intelligence'), affiliation types 'education', 'company', 'government'), improving scope accuracy resulting data. We evaluated approach against alternative solutions AIDA, MAG, four other benchmarks (FB15k, FB15k-237, WN18, WN18RR). outperforms when using low dimensions obtains competitive results in high dimensions.