Neo4j Graph Data Science has released version 2.3, which includes new algorithms such as a Knowledge Graph Embedding (HashGNN) for generating embeddings on heterogeneous graphs, improved performance and integration with various graph algorithms, undirected relationship types support, and enhancements to machine learning predictions on imbalanced or skewed datasets. The release also introduces common data science community datasets and provides a new procedure for writing labels based on node properties. Additionally, Cypher Aggregation is being targeted as the primary surface for projecting graphs from the database using Cypher, while Shard Local Algorithm Execution With Composite Databases allows for algorithm execution to be coordinated across multiple shards. The next major release will include fully managed graph data science as a service, AuraDS, on Amazon Web Services and Microsoft's Azure Platform.