Creating a projected graph with Cypher projection

In our auction fraud case, there is no direct relationship between users, but we still need to create a graph to run a centrality algorithm on. GDS offers a solution whereby we can create a projected graph for such cases using Cypher projection for nodes and/or relationships.

To create fake relationships between users, we consider them connected if they have joined at least one sale together. The following Cypher query returns these users:

MATCH (u:User)-[]->(p:Product)<-[]-(v:User) 
RETURN u.id as source, v.id as target, count(p) as weight

The count aggregate will be used to assign a weight to each relationship: the more common sales they have, the stronger the relationship between two users.

The syntax to create a projected graph using Cypher is as follows:

CALL gds.graph.create.cypher(
    "projected_graph_cypher", 
    "MATCH (u:User) 
        RETURN id(u) as id", 
    "MATCH (u:User)-[]->(p:Product)&lt...