Path finding from a Relation

[Yuval]

Is it possible to to start a path search from a relation rather than a node? 

The way we modeled are data, it turns out we have several use cases where the user chooses the relation rather than the node. 

[Wes Freeman]

You can do something like this:

start r=relationship(3) 

match p=n-[r]-()-[:KNOWS]-m 

return p;

http://console.neo4j.org/r/xex08r

Is that what you mean?

Wes

On Tue, Oct 2, 2012 at 7:01 PM, Yuval <yuval...@gmail.com> wrote:

Is it possible to to start a path search from a relation rather than a node? 

The way we modeled are data, it turns out we have several use cases where the user chooses the relation rather than the node. 

--
 
 

[Yuval]

Might very well be but I will need some assistance deciphering your answer (pardon the newbiness :-)

[Wes Freeman]

Commented example!

// given a known relationship -- can also do index lookup here

START r=relationship(3)  

// find all paths that have that relationship and then a :KNOWS relationship 

// (just an example--if you want something closer to your real use case, set it up in console and we can take a look)

MATCH p=()-[r]-()-[:KNOWS]-() 

// return those paths

RETURN p;

On Tue, Oct 2, 2012 at 7:14 PM, Yuval <yuval...@gmail.com> wrote:

Might very well be but I will need some assistance deciphering your answer (pardon the newbiness :-)

--

[Yuval Perlov]

Okay, i think i get it. I've been using the java API directly till now. Does it translate?  For these types of queries, should i rather use cypher?

--
 
 

[Wes Freeman]

I only know Cypher myself--am clueless about most of the APIs and stuff like Gremlin, and I've only been looking at graph databases for a couple of months, so take my opinion with a grain of salt. So, with that disclaimer, my opinion is biased, but I think you should use Cypher!

Anyone else?

In case you hadn't seen it, a quick example of Cypher through Java:

http://docs.neo4j.org/chunked/snapshot/tutorials-cypher-java.html

--
 
 

[Peter Neubauer]

Yves,

I would say Cypher is far more productive and readable then the Java API. However, if you need the last bit of fine grained control, then either the core API (iterators) or the Traversal API are your friends, since right now they provide more low-level access to the traversals than Cypher.

/peter

--
 
 

--

Cheers,

/peter neubauer

G:  neubauer.peter
S:  peter.neubauer
P:  +46 704 106975
L:   http://www.linkedin.com/in/neubauer
T:   @peterneubauer

Wanna learn something new? Come to http://graphconnect.com

[Eugeny Kozhanov]

It's my way:

1) set to start relationship flag "start"

2) finding path from one node with RelTypeFilterExcluder({"start"}) (it's your class which ignored "start" relationship, for example: http://vinger4.blogspot.com/2012/10/neo4j-class-reltypefilterexcluder.html)

3) finding path from other node with same RelTypeFilterExcluder.

4) join two results