Virtual Nodes and Relationships
Virtual Nodes and Relationships do not exist in the graph, they are only returned by a query, and can be used to represent a graph projection.
They can be used to visually project data, for example aggregating relationships into one, or collapsing intermediate nodes into virtual relationships. We could project a citation graph into a virtual author-author or paper-paper graph with aggregated relationships between them, or even turn Twitter data into a user-user mention graph.
We can combine real and virtual entities, for example by creating a virtual relationship between two real nodes or creating a virtual relationship from a virtual node to a real node.
Returns virtual entities without some defined properties
This section includes:
We can filter some properties of nodes and relationships present in a subgraph using the apoc.graph.filterProperties procedure,
or the analogous aggregation function.
For example, if we want to exclude embedding properties created with the
CALL apoc.ml.openai.embedding(["Test"], "<apiKey>", {}) yield embedding
with embedding
match (start:Start {id: 1}), (end:End {id: 2})
WITH start, end, embedding
CALL db.create.setNodeVectorProperty(start, "embeddingStart", embedding)
CALL db.create.setNodeVectorProperty(end, "embeddingEnd", embedding)
RETURN start, endwe would return virtual entities without those properties.
If we return the nodes to Neo4j Browser or Neo4j Bloom we would have the following situations, where we can se the log embedding properties :
But if we filter the embedding properties, then the situation would be as follows, easier to read:
