<ul><li>Knowledge graph completion aims to identify additional facts in the KG.</li><li>Recent developments in this field have explored KG completion in the inductive setting.</li><li>The CBLiP model introduces connection-biased attention and entity role embeddings to improve performance.</li><li>CBLiP outperforms models that do not use path information and is faster than models that utilize path information.</li></ul>

Replacing Paths with Connection-Biased Attention for Knowledge Graph Completion

Discover more