Feature Engineering from bipartite graph

Neo4j Graph Platform
1

I'm building a knowledge graph for customer and their attributes.
each customer is related to other type of node entity, but there is no relation between customer. I build bipartite graph.
here is the relation :
(customer) - [:BELONGS_TO] -> (Class)
(customer) - [:USING] -> (Power (VA))
(customer) - [:LIVES_IN] -> (subregion)
(customer) - [:HAVING_ARREARS) -> (Month)
I want to use the KG for feature engineering.
I tried Jaccard Similarity, but most of the score is 1.

Do anyone have suggestion what should I do with my graph?

2

Can you provide your query?

3

I'm using Graph Data Science Playground for the Jaccard similarity.
I'm using "Customer" as the label and ANY relationships as the graph projection