WalkLM: A Uniform Language Model Fine-tuning Framework for Attributed Graph Embedding

We conduct extensive experiments on two new real-world KG datasets, i.e., Freebase The nodes and edges are extracted according to [5]. A large portion of books are labeled into eight genres of literature. Each labeled book has only one label. FB15K-237 is a standard dataset in the knowledge graph community, which contains 310,116 triples with 14,541 entities and 237 relation types. Since we did not manually label the nodes, we only predicted whether a triple is correct or not on this dataset.