Hi Andrew'
It's something useful to play with :-)
Meantime I have started to get it work in javascript function.
May I know if you have faced to performance problems? My friends told me
that it was not easy to sort it out with gremlin on a bitcoin database. How
much data you deal with?
I have approx:
- vA: 1 million
- vB: 2 million
- vC: around 200-300 million
- vD: 2 million
- vE: 200 k
And having few other class of vertices, 1 million each.
I think it still not "big data", but something I can manage on a commodity
oracle server. But it is not good for realtime recommender engine, although
the performance is pretty good :-)
Thx!
L.
2014.03.27. 14:03 ezt írta ("Andrey Yesyev" <[email protected]>):
> Hi László,
>
> I found this site is very useful http://gremlindocs.com/
>
> This is how I look for all paths between 2 vertices, with rids #21:7 and
> #16:10
>
> g.v('#21:7').as('person').both('RELATED').loop('person'){it.object.id.toString()
>> != '#16:10' && it.loops < 4}.filter{it.id.toString()=='#16:10'}.path()
>
>
> This is how java code looks like
>
> OrientGraph graph =
>> DatabaseManager.getInstance().getDatabase("remote:
>> 192.168.200.128/ePersona").getConnection();
>> Pipe pipe =
>> Gremlin.compile("_().as('person').both('RELATED').loop('person'){it.object.id.toString()
>> != '#16:10' && it.loops <
>> 4}.filter{it.id.toString()=='#16:10'}.path()[10..20]");
>> pipe.setStarts(new
>> SingleIterator<Vertex>(graph.getVertex("#21:7")));
>> Merger m = new Merger();
>> Path paths = m.mergePathsResults(pipe);
>> for(Object o : pipe) {
>> ArrayList<Vertex> path = (ArrayList<Vertex>)o;
>> for(Vertex v:path){
>> System.out.print("->" + v.getProperty("name") + "[" +
>> v.getProperty("@class") + "]");
>> }
>> System.out.println();
>> }
>
>
> example with weights you can find on gremlindocs site.
>
> -Andrey
>
>
> On 27 March 2014 01:44, László Rékasi <[email protected]> wrote:
>
>> Hi,
>>
>> Thank you!
>> Thought that it could not be easy :-) Meantime realized that there are
>> nodes and egdes in chain :-) but not projected result sets like in rdbms
>> world.
>>
>> Unfortunately I am not a developer, but a guy in business side with some
>> it skill, so somehow I need to make it with out of the box solution to
>> proof that it worth to spend time on it. So maybe javascript function can
>> be the answer somehow.
>>
>> @Andrey, could you please share your use case and solution with me?
>> Orientdb lacks of complete, step by step tutorial, or at least I did not
>> find too many.
>>
>> Thx!
>> L.
>> 2014.03.27. 2:08 ezt írta ("Andrey Yesyev" <[email protected]>):
>>
>> Hi there,
>>>
>>> I have the similar problems. I solved them using Gremlin-Groovy.
>>> I doubt you can do what you need using SQL.
>>>
>>>
>>> -Andrey
>>>
>>> On Tuesday, March 25, 2014 3:58:06 PM UTC-4, László Rékasi wrote:
>>>>
>>>> Hi,
>>>>
>>>> I have the following datastructure (every entity has its own class):
>>>> vertexes : vA, vB, vC, vD, vE
>>>> there is eAB between vA and vB with one to many relation
>>>> eBC between vB and vC with one to many relation
>>>> eDC between vD and vC with one to many relation from D to C (so vC is a
>>>> for resolving many to many between vB and vD, but having lots of
>>>> properties)
>>>> eED between vE and vD with one to many relation from E to D
>>>>
>>>> Now I would like to find vA and vD nodes where there is a path between
>>>> with the following conditions and expectations:
>>>>
>>>> - I need the ID and some fields from both vA and vD
>>>> - want to know the number of path between vA and vD
>>>> - want to sum() some fields in vC as weight
>>>> - want to filter on fields of all vertexes (vA.fld='X', vB.fld_b =
>>>> 'Y', etc
>>>>
>>>> Basically the situation is something like:
>>>> [person] - (having)->[user accounts]
>>>> [user account]-(post)->[message]-(to)->[recepients' account]
>>>> [recepient's account]-(belongs to)->[user]
>>>>
>>>> So want to know who sent messages in certain time to X type of user and
>>>> how many times. Wants to know the first and last date. The sender belongs
>>>> to Y domain.
>>>> person.domain = 'Y'
>>>> message.sent_date between D1 and D2
>>>> user.type = 'hacker'
>>>>
>>>> I need it in SQL.
>>>>
>>>> Any suggestion is very appreciated!
>>>> Thank you in advance!
>>>> L.
>>>>
>>> --
>>>
>>> ---
>>> You received this message because you are subscribed to a topic in the
>>> Google Groups "OrientDB" group.
>>> To unsubscribe from this topic, visit
>>> https://groups.google.com/d/topic/orient-database/IKbefI2QA3M/unsubscribe
>>> .
>>> To unsubscribe from this group and all its topics, send an email to
>>> [email protected].
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>> --
>>
>> ---
>> You received this message because you are subscribed to a topic in the
>> Google Groups "OrientDB" group.
>> To unsubscribe from this topic, visit
>> https://groups.google.com/d/topic/orient-database/IKbefI2QA3M/unsubscribe
>> .
>> To unsubscribe from this group and all its topics, send an email to
>> [email protected].
>> For more options, visit https://groups.google.com/d/optout.
>>
>
> --
>
> ---
> You received this message because you are subscribed to a topic in the
> Google Groups "OrientDB" group.
> To unsubscribe from this topic, visit
> https://groups.google.com/d/topic/orient-database/IKbefI2QA3M/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> [email protected].
> For more options, visit https://groups.google.com/d/optout.
>
--
---
You received this message because you are subscribed to the Google Groups
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
