Hey all!
I'm just getting started with Neo4j and it's been super fun, but I've hit a
wall I don't know how to navigate around - here's the gist of what I'm
trying to do:
I have networks, which have many "egos" that appear in those networks.
These egos, in turn, occur in many different networks - basic many-to-many
relationship, and in the context of a graph db, a bipartite network. In
practice, my data looks something like this (pardon whatever silly format
I'm about to write it as):
Network 1: [Ego 1, Ego 2, Ego, 3, Ego 4]
Network 2: [Ego 2, Ego 4, Ego, 5, Ego 6]
Network 3: [Ego 4, Ego 5, Ego, 7, Ego 8, Ego 9, Ego 10]
What I would like *out* of this data is something like the following:
I want to know the relationship that Network 1 has with Networks 2 and 3.
Specifically, I want to measure the overlap in a specific way - the count
of the intersection of egos for two networks divided by the count of the
union of egos. In other words, a comparison of Network 1 and Network 2
would yield the following steps:
intersection = [Ego 2, Ego 4], 2 total
union = [Ego 1, Ego 2, Ego 3, Ego 4, Ego 5, Ego 6], 6 total
intersection/union = 1/3
I would then say that these two networks are 33% similar (I'm open to other
ways to calculate similarity in the spirit of what I'm trying to do here).
For reference, the query I've got right now can pull out the intersection
but not the union - also, any hints/tips about indexing/cypher query
improvement/server optimization is very welcome.
The query in question:
MATCH (network:Network {id:
'52acc273bf34801f89068a5a'})-[:appearance]->(ego:Ego)-[:occurrence]->(alter_network:Network)
return distinct(alter_network.id), count(ego)
Thanks much! Loving my experience so far,
Devin
--
You received this message because you are subscribed to the Google Groups
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
