Text this: Towards leveraging explicit negative statements in knowledge graph embeddings