(Un)likelihood Training for Interpretable Embedding - Citation Graph | Papersgraph