CODEWITHSUNDEEP
neo4jcypher
skme
skme

Reputation: 761

cypher count different permutation of connections between nodes

I have a search items graph model in neo4j.

So I have the following data types nodes: Item , Key, and category

I want to do a statistic check on the graph. I want to know if I take one key out of each category - how many connected Items there are for every set of keys.

So, if I had key_A(12) from category_A and key_B(34) from Category_B, I could do:

START key1 = node(12), key2 = node(34)
MATCH key1 <--(item), key2<--(item)
RETURN count(item)

Pretty simple... But I need to run the same for every set of keys. (one key from each category)

How can I run the same question for each permutation of the keys?

Upvotes: 0

Views: 343

Answers (2)

Michael Hunger
Michael Hunger

Reputation: 41676

Can you try this?

start cat=node:node_auto_index("name:(Category1 Category2 Category3")
MATCH (item)-->(key)<--(cat)
where (item.type="item")
with key.name as key_name, collect(distinct item.name) as items
order by key.name
with collect(key_name) as keys, reduce(a=[],i in collect(items) : a + i) as items2
return keys, items2, length(items2)

Probably posting an example dataset to http://console.neo4j.org

or describing it as a graph-gist (http://gist.neo4j.org) would be great.

Upvotes: 0

skme
skme

Reputation: 761

Found the answer!

start cat1=node:node_auto_index(name="Category1"),cat2=node:node_auto_index(name="Category2")
MATCH (item) --> (key1)<--(cat1),
      (item)--> (key2)<--(cat2)
where (item.type="item")
with key1, key2 , collect(distinct item.name) as items    
return key1.name,key2.name, items, length(items)

Upvotes: 2

Related Questions