Andreas Kollegger wrote: > Oh, and here's the graph... > > > > On Tue, Sep 7, 2010 at 10:59 AM, Andreas Kollegger < > [email protected]> wrote: > > >> Hey Stephane, >> >> It occurs to me that a general strategy for improving traversal speeds is >> to add indexing nodes that help reduce the number of possible paths to >> check. For instance, you could add "week of the year" nodes which group bank >> receipts and statements into a time-oriented groups. In the attached graph, >> I've taken your last example and added a "date reference" sub-reference >> node, and then wk1 and wk2 "week of the year" nodes to group the occurrences >> of receipts and statements. >> >> This approach can be elaborated to greatly reduce the number of edges to >> iterate over and nodes to check. Think of your trees as tries ( >> http://en.wikipedia.org/wiki/Trie). >> >> Best, >> Andreas >> >> On Mon, Sep 6, 2010 at 8:45 PM, Stephane Urdy < >> [email protected]> wrote: >> >> >>> Peter Neubauer wrote: >>> >>>> Stephane, >>>> what kind of property or pattern are you matching on to establish the >>>> relationship between BR and BS? >>>> >>>> Cheers, >>>> >>>> /peter neubauer >>>> >>>> COO and Sales, Neo Technology >>>> >>>> GTalk: neubauer.peter >>>> Skype peter.neubauer >>>> Phone +46 704 106975 >>>> LinkedIn http://www.linkedin.com/in/neubauer >>>> Twitter http://twitter.com/peterneubauer >>>> >>>> http://www.neo4j.org - Your high performance graph >>>> >>> database. >>> >>>> http://www.thoughtmade.com - Scandinavia's coolest Bring-a-Thing party. >>>> >>>> >>>> >>>> On Wed, Sep 1, 2010 at 8:19 PM, Stephane Urdy >>>> <[email protected]> wrote: >>>> >>>> >>>>> Hi, >>>>> >>>>> I was searching on forums how to match nodes with other nodes when I >>>>> >>> saw >>> >>>>> a message where you mentioned patterns. >>>>> >>>>> I am very new to the world of graph databases and if you have a little >>>>> time I would be glad if you could answer this question. >>>>> >>>>> I would like to match receits nodes with bank accounts statement entry >>>>> nodes. >>>>> in the 2 examples, br2 [bank receit number2] matches bs1 [bank >>>>> >>> statement >>> >>>>> entry number1] >>>>> >>>>> What would be the most efficient model to match the nodes ? >>>>> example1.png shows 2 trees with one level >>>>> example2.png shows 2 lists where each entry has a next entry [here my >>>>> undestanding that we exploit the tranversal speed] >>>>> >>>>> The basic idea is to take the first receit node and compare it to each >>>>> bank receit entry node, then add the relationship between the 2 >>>>> >>> matching >>> >>>>> nodes [br--matches-->bs] >>>>> >>>>> I could possibly add an index in the equation to speed things up, but >>>>> basically my goal is to find the most efficient ways to doing it. >>>>> >>>>> Any pointers would be appreciated, patterns included :) . >>>>> >>>>> Kind regards, >>>>> >>>>> Stephane >>>>> >>>>> >>>>> >>>>> >>>>> >>> ------------------------------------------------------------------------ >>> >>>>> >>>>> >>> ------------------------------------------------------------------------ >>> >>>>> _______________________________________________ >>>>> Neo4j mailing list >>>>> [email protected] >>>>> https://lists.neo4j.org/mailman/listinfo/user >>>>> >>>>> >>>>> >>>>> >>>> _______________________________________________ >>>> Neo4j mailing list >>>> [email protected] >>>> https://lists.neo4j.org/mailman/listinfo/user >>>> file:///home/stef/Desktop/example3.png >>>> >>>> >>>> >>> Hi Peter, >>> >>> I will probably match br and bs entries with a few attributes common to >>> both bs entries nodes and receits entries nodes: >>> >>> - the account name >>> - the amount >>> - Whether it is a debit or a credit >>> - the date more or less n days >>> - the type ex: direct debit,transfer,cash machine withdrawal,standing >>> order, swift payment, bank credit ... >>> >>> I finally started to model using a tree structure with 3 reference nodes >>> : one for the accounts, one for the bs entries and one for the receits >>> entries. >>> >>> not so sure it is the most appropriate way to go for fast matching my >>> entries. >>> I added a visual in the attached file example3.png so you can have an >>> idea once the model is populated with a small amount of entries/nodes. >>> >>> Any better idea are welcomed. >>> >>> Thank you for your support and expertise, >>> >>> Cheers, >>> >>> Stephane >>> >>> _______________________________________________ >>> Neo4j mailing list >>> [email protected] >>> https://lists.neo4j.org/mailman/listinfo/user >>> >>> >>> >> >> ------------------------------------------------------------------------ >> >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> Neo4j mailing list >> [email protected] >> https://lists.neo4j.org/mailman/listinfo/user >> Hi Andreas,
It seems like a good approach, I will try this. Many thanks for your input. Kind regards, Stephane _______________________________________________ Neo4j mailing list [email protected] https://lists.neo4j.org/mailman/listinfo/user
