Get all paths between two nodes via specific node

Neo4j Graph Platform
1

We have a Neo4J database that tracks company Supply for companies. The general structure is

````````(Company)-[:Supply_For]->(Company)`

A company can have multiple rels at the same time and each company can have multiple company supply rels.

my question is I want to find the path via two given nodes id, I tried as below pattern:

``match p = (c1:Company{id: "abc"})-[:Supply_FOR*0..2]->(c) <- [:GLF*0..2]-(c2:Company{id: "def"}) return p

but the Response Time cannot accept, how can I get this requirement?

2

Hi @Reid
Did you get any chance to try apoc path finding algorithms (https://neo4j.com/labs/apoc/4.4/algorithms/path-finding-procedures/)? Btw, what is the issue with the Cypher path finding method that you are trying to use? Does it take a lot of time? If so, can you provide with some more information about the amount of nodes, rels, model and the configuration of Neo4j?

All the best

3

thank you for your reply.

I read the apoc path finding algorithms, didn't find which algo meets my requirement.

I want is A - [Supply_FOR*0..2] ->(c)<-[Supply_FOR*0..2]-(B)

it's more like finding a path between A and B and here sometime allowed another one node C,

In general, three patterns including:

1. A -[Supply_FOR*1..4] -> (B)

2. B - [Supply_FOR*1,,4] -> (A)

  1. A - [Supply_FOR*1..2] ->(c)<-[Supply_FOR*1..2]-(B)

---

The Cypher I use is

``match p = (c1:Company{id: "abc"})-[:Supply_FOR*0..2]->(c) <- [:Supply_FOR*0..2]-(c2:Company{id: "def"}) return p limit 20`

it's too slow, many requests (30%) cost over 10 seconds.

nodes: 586 million, rels: 982 million

BTW, nodes I need to find of A and B, maybe a super node.

neo4j info:

Version:

4.4.3

Edition:

Community

Name:

neo4j

4

is there any suggestion?

5

Hi,
Thanks for sharing some more information about the neo4j configuration. What about the heap_size and pagecache allocation? Because the graph that you have is considered to be a big one, and it is running on Neo4j Community not on Enterprise.

Regarding the APOC you could try the following:

MATCH (c1: Company {id: "abc"})
CALL apoc.path.expandConfig(c1, { relationshipFilter: "Supply_FOR", minLevel: 0, maxLevel: 2 }) 
YIELD path
RETURN path
LIMIT 20

And then you can use the same approach to find all possible paths, while including the second/parent Company as well.

6

heap_size 8G， dbms.memory.pagecache.size

I don't think this APOC meet my requirement.

Firstly, I want to find paths between A and B, this statement just have one node A.

Secondly, all paths the direction is specific.

7

Hi @Reid

The query below:

MATCH (c1: Company {id: "abc"})
CALL apoc.path.expandConfig(c1, { 
    relationshipFilter: "<Supply_FOR>",
    labelFilter: "+Company",
    filterStartNode: true, 
    uniqueness: "RELATIONSHIP_PATH", 
    bfs: false
}) 
YIELD path
RETURN path

Starts at the Company {id: "abc"} and it expands all relationships of a type Supply_FOR in all directions (either incoming or outgoing) and it gets the entire path until the last leaf node. So basically performs a depth-first-search algorithm while keeping the uniqueness of the relationship type. If this does not satisfy what you want to do, maybe you could revise your model or the requirements. I hope that this answer helps you.

Regards,