Finding Interpretable Concept Spaces in Node Embeddings using Knowledge Bases