Comparing Nodes in two separate graphs for Similarity

Neo4j Graph Platform Graph Algorithms/Graph Data Science
, ,
1

Hello Everyone,

I'm a student trying to compare two separate singular nodes based on their neighbors in separate graphs with each other.
If I understood correctly, the node similarity algorithms only work within a single graph; how could I get around that?
The graphs I'm trying to compare should generally be very similar to each other, and I just need to find out if/where any differences would be.

Thanks for your help in advance.

2

Do you have any node properties?
You could look into generating node embeddings, such as with FastRP and a property ratio of 1 (see Fast Random Projection - Neo4j Graph Data Science).
As your nodes are in distinct graphs, you want to look for inductive algorithms.

Next, you compute the similarity between these embeddings (such as with K-Nearest Neighbors - Neo4j Graph Data Science).

Hope these pointers help.

3

I think node embeddings might be just what i need, but i'm unsure as to how to apply the similarity algorithms onto the embeddings.
Also thank you very much for the fast reply.

4

Hey @stefan.grad ,
you can use the nodeProperties parameter of KNN to specify the property containing your previously computed embedding. (parameter example)