If we want to recommend new repositories to our GitHub users, we can think of the following rule: repositories of a user I follow are more likely to be of interest to me, otherwise I wouldn't follow those users. We can use Cypher to identify those repositories:
MATCH (u:User {login: "mkhq"})-[:FOLLOWS]->(following:User)-[:CONTRIBUTED_TO]->(recommendation:Repository)
WHERE NOT EXISTS ((u)-[:CONTRIBUTED_TO]->(recommendation))
RETURN DISTINCT recommendation
This query matches patterns similar to the following one:
We can also use recommendation ordering here. The higher the number of people I follow that also contributed to a given repository, the higher the probability that I will also contribute to it. This translates into Cypher in the following way:
MATCH (u:User {login: "mkhq"})-[:FOLLOWS]->(following:User)-[:CONTRIBUTED_TO]->(recommendation:Repository)
WHERE NOT EXISTS ((u)-[:CONTRIBUTED_TO]->(recommendation...
