List the nodes forming unrelated clusters

[masavini]

Hi everybody!

I have a graph with a few thousand nodes which form unrelated clusters. I'd like to get a list of lists with the nodes with a certain label forming each cluster, regardless of the relation type linking them.

Let's consider this simple example:

CREATE
    (:Person {name: "Andrea"})-[:LOVES]->(:Person {name: "Bob"})-[:LIKES]->(:Food {name: "pizza"}),
    (n:Person {name: "Mike"})<-[:KNOWS]-(:Person {name: "Paul"})-[:KNOWS]->(:Person {name: "Jen"})<-[:LOVES]-(n)

If I'd like to know which persons are somehow related, then I would get:

[
    ["Andrea", "Bob"],
    ["Mike", "Paul", "Jen"]
]

Could please help me building such a query? Thanks!

[masavini · Accepted solution]

Thank you very much, that is exactly what I was looking for.

So I first projected a new graph:

CALL gds.graph.project(
  'myGraph',
  'Person',
  '*'
)

Then I retrieved information about each cluster (aka component) in myGraph:

CALL gds.wcc.stream('myGraph')
YIELD nodeId, componentId
RETURN gds.util.asNode(nodeId).name AS name, componentId
ORDER BY componentId, name

╒════════╤═════════════╕
│"name"  │"componentId"│
╞════════╪═════════════╡
│"Andrea"│0            │
├────────┼─────────────┤
│"Bob"   │0            │
├────────┼─────────────┤
│"Jen"   │2            │
├────────┼─────────────┤
│"Mike"  │2            │
├────────┼─────────────┤
│"Paul"  │2            │
└────────┴─────────────┘

Could you suggest how to group names inside a list of lists?

[cuneyttyler]

You might need to extract connected components, Check https://neo4j.com/docs/graph-data-science/current/algorithms/wcc/

[masavini · Accepted solution]

Thank you very much, that is exactly what I was looking for.

So I first projected a new graph:

CALL gds.graph.project(
  'myGraph',
  'Person',
  '*'
)

Then I retrieved information about each cluster (aka component) in myGraph:

CALL gds.wcc.stream('myGraph')
YIELD nodeId, componentId
RETURN gds.util.asNode(nodeId).name AS name, componentId
ORDER BY componentId, name

╒════════╤═════════════╕
│"name"  │"componentId"│
╞════════╪═════════════╡
│"Andrea"│0            │
├────────┼─────────────┤
│"Bob"   │0            │
├────────┼─────────────┤
│"Jen"   │2            │
├────────┼─────────────┤
│"Mike"  │2            │
├────────┼─────────────┤
│"Paul"  │2            │
└────────┴─────────────┘

Could you suggest how to group names inside a list of lists?

[cuneyttyler]

Can you give an example about 'list of lists'. What kind of data is this?

[masavini]

in the example above, something like this:

[
    ["Andrea", "Bob"],
    ["Mike", "Paul", "Jen"]
]

a list of components where each component is represented by a list of node names.

[cuneyttyler]

In neo4j there is no explicit grouping logic. Group by works with aggregate functions so I will use 'count'. I suppose something similar to this would work for you.

WITH [
    ["Andrea", "Bob", "Andrea"],
    ["Mike", "Paul", "Jen","Bob"]
] as myList UNWIND myList as subList UNWIND subList as name return name, count(name)

returns

"Andrea"	2
"Bob"	2
"Mike"	1
"Paul"	1
"Jen"	1

[glilienfield · Accepted solution]

This groups the names by componentId.  Do you want further grouping?

CALL gds.wcc.stream('myGraph')
YIELD nodeId, componentId
RETURN collect(gds.util.asNode(nodeId).name) AS names, componentId
ORDER BY componentId

[masavini]

what can i say? OUTSTANDING job, guys. never seen such a good support before, congratulations.