- From: Graham Klyne <gk@ninebynine.org>
- Date: Wed, 24 Mar 2004 19:18:34 +0000
- To: Tim Berners-Lee <timbl@w3.org>
- Cc: www-rdf-calendar@w3.org
Thanks! When I have some time, I might have a look at generating patch files from my graph-diff. (I *think* I can be sure of not generating falsehoods in this way, at the expense of a somewhat "brute-force" approach to finding possible node-matches after the obvious ones have been isolated.) #g -- At 11:10 24/03/04 -0500, Tim Berners-Lee wrote: >Ooops. Sorry, No, Graham, [1] should have been to "Diff, patch, update, >sync" >[1] http://www.w3.org/DesignIssues/Diff >which points to an actual program >[3] http://www.w3.org/2000/10/swap/diff.py > >The algorithm in diff.py is to find a mapping corresponding nodes in the >two graphs >because they have the same URI or literal value or they are connected by >an [inverse] >functional property to nodes which have already been mapped as corresponding. > >It isn't the pretty-printing-like algorithm which you use in [2]. >It only works on a subset of graphs which have are well labelled, but the >resulting >diff file has useful property that you can apply it to anything without >fear of >generating falsehoods. > >The update language looks like for example >[4] http://www.w3.org/2000/10/swap/test/delta/t2/patch.n3 > > >Tim > >On Mar 24, 2004, at 6:45, Graham Klyne wrote: > >>At 16:36 23/03/04 -0500, you wrote: >>>To make the graph labelled enough for doing a diff,[1], with calendar >>>data, here are the fixes I had to make (a copy of) to the ontology: >> >>I didn't see a reference given for '[1]'. If it's my graph-differencing >>function [2], I'd be interested to see the problem data, as my algorithm >>doesn't depend *entirely* on graph labels. If not mine, I'd be >>interested to know of other approaches. >> >>#g >>-- >> >>[2] http://www.ninebynine.org/RDFNotes/Swish/Intro.html#GraphDiff >> >> >>------------ >>Graham Klyne >>For email: >>http://www.ninebynine.org/#Contact > >------------ >Graham Klyne >For email: >http://www.ninebynine.org/#Contact
Received on Wednesday, 24 March 2004 16:22:24 UTC