Graphs - Python - Store a graph db made of graphs - Graph db theory

Question

I'm building an application that sits on a graph data set, and nodes in this graph can be other smaller graphs, in wich the nodes can have connections to the main graph's nodes.

A (badly drawn, sorry :D ) ASCII art to compensate my (bad, sorry :D) english:

               //       \\BAR
              ||(x)---(y)||
              ||      /  ||\
              ||    (z)  ||\\
               \\    |  //  \\
              //     |       \\_________
             //      |                  \
            //       |             //    \           \\
FOO        //        |             ||(harry)---(terry)||
 //       \\         |             ||          /      ||WORLD
||(a)---(b)||        |            /||    (jerry)      ||
||  \      ||        |           // \\                //
||   (c)   ||       /  //      \\/
 \\   |   //       /  || HELLO ||
      \___________/    \\     //

Node x,y and z are nodes of the BAR graph, a,b and c of the FOO graph, same for harry ecc.. and the WORLD graph.

FOO, BAR, WORLD and HELLO are nodes of the main big Graph.

Inside this structure a node of FOO is connected to a node in BAR (c -> z) and, this is the tricky part) a node of WORLD is connected to BAR (harry -> BAR).

I know how to achieve this with Networkx in python, but how can I save this to a db?

I'm thinking of solutions myself but I want to know if there is a theory or some techniques for similar situations.

So far my solution (using neo4j) is label a b and c with the tag "FOO", and making a separate graph that connects FOO to BAR, ecc.. Am I in the right direction? Do you know some smarter solutions?

Thanks

[edit: corrected a name]

cybersam · Accepted Answer · 2014-09-25 17:52:57Z

1

You did not label "J" in your diagram, so I will ignore it (and HELLO, whatever that is) in this answer.

[EDITED]

You just need a single node instance of each subgraph, and they can all have the label Subgraph. I'll name these nodes foo, bar, and world below.

(foo:Subgraph {name: "Foo"})
(bar:Subgraph {name: "Bar"})
(world:Subgraph {name: "World"})

Then you can indicate which nodes are members of which subgraph:

(foo)-[:MEMBER]->(a)
(foo)-[:MEMBER]->(b)
(foo)-[:MEMBER]->(c)

(bar)-[:MEMBER]->(x)
(bar)-[:MEMBER]->(y)
(bar)-[:MEMBER]->(z)

(world)-[:MEMBER]->(harry)
(world)-[:MEMBER]->(terry)
(world)-[:MEMBER]->(jerry)

And you can also have relationships that connect member nodes as wells as entire subgraphs:

(c)-[rel1]->(z)
(harry)-[rel2]->(bar)

edited Sep 25, 2014 at 17:52

answered Sep 25, 2014 at 16:30

cybersam

67.3k6 gold badges59 silver badges81 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Hrabal Over a year ago

Yeah sorry I changed J with HELLO but not in the text... - With this solution the labels do not become superfluous? I'm thinking of labeling my subgraphs nodes as Subgraph so I can query only the connections between subgraphs (MATCH (node1:Subgraph)-[:rel1]->(node2:Subgraph)).. Is this correct?

cybersam Over a year ago

Yes, you are right. And if you still wanted to name each subgraph, you could just have a name property in each Subgraph instance. I will edit my answer.

Collectives™ on Stack Overflow

Graphs - Python - Store a graph db made of graphs - Graph db theory

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related