Hi Michael,
It is fun to see you thinking this through :-)
I have had a chance to discuss my crazy ideas with some of the neo4j team
last year, but this is the first time with you. And since December there is
even code to review (see https://github.com/craigtaverner/amanzi-index).
You are
Tim,
I don't know enough about disambiguation algos to point you in any
special direction, but implementation-wise you could hook into the
event framework, http://wiki.neo4j.org/content/Event_framework, in
order to do e.g. graph matching queries upon any modifying transaction
changing database
I was working on a project that used matching algorithms a while back.
What you have is an n-dimensional matching problem. I can't remember
specifically what the last project were using, but this and the linked algos
may be what you're looking for:
Here is a crazy idea - how about taking the properties you care about and
dropping them into a combined lucene index? Then all results for nodes with
the same properties would be 'ambiguous'. Moving this forward to degrees of
ambiguity might be possible by creating the combined 'value' using a
Craig,
how do you map _all_ properties to the integer (or rather numeric/long?) space.
Does the mapping then also rely on the alphabetic (or comparision) order of the
properties?
Interesting approach. Is your space then as n-dimensional as the numbers of
properties you have?
Cheers
Michael
Thanks so much for that explanation.
Put it in a blog post.
Michael
Am 02.02.2011 um 01:33 schrieb Craig Taverner:
Hi Michael,
I agree that _all_ is a strong work :-)
What I do is provide a mapper interface for the domain modeler to implement
based on their own understanding of what
Craig,
The interesting thing is - if you do this index in-graph you can directly
attach the
index-nodes to the real nodes and have them mimic the original node but instead
of having
the property value per property you have the mapped value.
And you can many of such indexes in graph attached
7 matches
Mail list logo
