Computing new node representations

Similarly to the previously studied algorithms, we can extract the results via streaming, writing to Neo4j, or mutating the projected graph. We will learn about the use case for the later mode in the next chapter (Chapter 8, Building a GDS Pipeline for Node Classification Model Training), but for now, let’s use the simple streaming mode:

graph_object = create_projected_graph(
     gds,
     "pgraphTest",
     node_spec="Test",
     relationship_spec={
           "RELATED_TO": {"orientation": "UNDIRECTED"}
     },
     nodeProperties=["x0"]
)
embeddings = gds.beta.graphSage.stream(
     graph_object,
     modelName="myGS&quot...