Hi,
> Now that looks interesting! But why limit it to a pipe and not merge
> it with the Graph Matching component, so we can find non-linear
> subgraphs with the same API? Or did I misunderstand the concept?
Graph matching can be seen as a subset of graph traversal. Meaning,
any graph match can be represented as a graph traversal, but not all
graph traversals can be represented as a graph match.
You might be interested in Pipes and Gremlin as a concise way to
express graph traversals and as such, express arbitrary graph matching:
http://pipes.tinkerpop.com
http://gremlin.tinkerpop.com
With Gremlin you can do graph matching that includes operations such
as union, intersect, deduplicate, math ops, equals, etc. In short, you
can conveniently apply any computable criteria to your definition of a
"graph pattern."
For those that care:
1. Unlike languages like SPARQL (1.0), you can express recursion and
thus, regular paths.
2. Unlike regular path languages, you are provided memory in your
traversal and thus, can express context-free paths (there is a notion
of history).
3. Unlike context-free path languages, you can brach, have arbitrary
memory and thus, can compute Turing complete paths.
Take care,
Marko.
http://markorodriguez.com
_______________________________________________
Neo4j mailing list
User@lists.neo4j.org
https://lists.neo4j.org/mailman/listinfo/user
