Graph Machine Learning Dataset SOA-SW

SOA-SW is a heterogeneous graph machine learning dataset based on the RDF knowledge graph SemOpenAlex-SemanticWeb. SOA-SW contains six node types - works (95,575 nodes), authors (19,970 nodes), concepts (38,050 nodes), sources (10,739 nodes), institutions (5,846 nodes), and publishers (786 nodes), and seven edge types. Each node has rich semantic node features as node representation (content-based and topology-based node features are available). More information can be found in the README.txt and on https://github.com/davidlamprecht/AutoRDF2GML. soa-sw-homogeneous-author only models the co-author network of SOA-SW. It is a homogeneous graph containing the author node type (19,970 nodes) and the edge type author- author. The authors' content-based features (nodes-nld) are based on the titles and abstracts of the authors' works (128-dimensional SciBERT embeddings).