Tag
The paper presents an inductive framework for learning language-independent concept embeddings from cross-linguistic colexification networks combined with textual concept descriptions, enabling embeddings for unseen concepts and correlating well with cross-linguistic similarity measures.