How to perform distinct while having multiple paths using Cypher - neo4j

Question

For a given node(sourceNode) I want to retrieve all node's which has relationships to my sourceNode within 3 hops.
The problem starts when we have multiple pathes btw source and destination nodes.
I dont care which path I get as long as I get one and I dont want to get the other ones (would be great to get only the shortest path)
So this is my code:
MATCH (user:C9 {userId:'70'})-[r:follow*1..3]-f WHERE f <> user
RETURN DISTINCT (f.userId) as userId,
reduce(s = '', rel IN r | s + rel.dist + ',') as dist,
length(r) as hop
The repose for this consist the same nodeId(userId's) and not performing distinct:
I would like to avoid the duplicated lines with the same userId.
any idea how to perform the distinct here?
Thanks,
ray.

Answer 1

How about something like this? Rather than look for distinct user just ue shortestPath to get to each follower 1..3 out from the starting user.
MATCH p=shortestPath((user:C9 {userId:'70'})-[r:follow*1..3]-(f))
WHERE f <> user
RETURN f.userId,
reduce(s = '', rel IN r | s + rel.dist + ',') as dist,
length(p) as hop
Alternatively, if you were looking to do it by shortest distance regardless of hops you could do something like the following example. Instead of using shortestPath, aggregate the distances on each relationship, order by shortest, put them in a collection, order by user and return the first element of the collection which will be the shortest
MATCH p=(user:C9 {userId:'70'})-[r:follow*1..3]-(f)
WHERE f <> user
with f.userId as user_id
, reduce(s = 0, rel IN relationships(p) | s + rel.dist) as dist
, length(p) as hops
order by dist
with user_id, collect(dist) as dists_per_follow, collect(hops) as hops_per_follow
return user_id
, dists_per_follow[0] as shortest
, dists_per_follow, hops_per_follow
order by user_id