- From: Patrick Stickler <patrick.stickler@nokia.com>
- Date: Wed, 14 Apr 2004 09:15:25 +0300
- To: "ext Chris Bizer" <chris@bizer.de>
- Cc: <www-archive@w3.org>, "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com>, "ext Pat Hayes" <phayes@ihmc.us>
On Apr 13, 2004, at 17:37, ext Chris Bizer wrote: > > Coming back to the graph/warrant cardinality question: > >>> I >>> think there are many situations, where you want to attach several >>> graphs to >>> one warrent, e.g. your are a information intermediary and you want to >>> say >>> that you quote all the 500 graphs you pass on. >> >> But how would you sign the warrant? >> >> Patrick > > Our signature method defines three things: > 1. The graph/graphset canonicalization method (e.g. what Jeremy > proposed in > his signing RDF paper) > 2. The hash function for hashing the canonicalized graph/grapset > 3. The Signature algorithm for signing the hash value. > > There is no problem in defining a canonicalization method for > graphsets, > thus it is also possible to sign them. > > In order to avoid unnecessary metadata, I still think we should loosen > the > cardinality between graph and warrant. Another argument is > compatibility, > XML Signature also allows signing several resources at once. I see your point. In any case, I don't find there to be any such 1:1 cardinality defined in the schemas, or in the comments, so I think we're OK with a warrant/certification being associated with more than one graph (not that I would ever expect it to happen in real usage). Patrick > > Chris > > ----- Original Message ----- > From: "Patrick Stickler" <patrick.stickler@nokia.com> > To: "ext Chris Bizer" <chris@bizer.de> > Cc: <www-archive@w3.org>; "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com>; > "ext > Pat Hayes" <phayes@ihmc.us> > Sent: Thursday, April 08, 2004 7:38 AM > Subject: Re: Warrent or PublishingEvent or Commitment and Cardinality > > >> >> On Apr 07, 2004, at 18:44, ext Chris Bizer wrote: >> >>> >>> Hi, >>> >>> >>> >>> I had a look at the swp.rdfs schema and >>> >>> >>> >>> 1. I'm thinking now that "PublishingEvent" is too restrictive. >>> Somebody >>> might name an publish a graph. Somebody else might quote it, a third >>> person >>> might also assert it ... So what about calling the thing >>> "Commitment", >>> a >>> term which is open for all kinds of relationsships, even others >>> beside >>> of >>> asserting and quoting. >> >> I'm really liking "Certification", (or else "Voucher"). >> >> I think commitment may suggest alot more legal machinery (or need for >> comprehensive explainations) than we want to bother with... >> >>> >>> >>> >>> 2. Patrick's comment in the schema defines the cardinality between a >>> "warrent" or whatever it is called and a graph as a one-to-one >>> relation. >> >> Given that a signature in a warrant/certification would be graph >> specific, >> I'm not sure how this relationship wouldn't be percieved to be >> one-to-one >> (not that I think the language of the comment necessarily states so >> strict >> a cardinality). >> >>> I >>> think there are many situations, where you want to attach several >>> graphs to >>> one warrent, e.g. your are a information intermediary and you want to >>> say >>> that you quote all the 500 graphs you pass on. >> >> But how would you sign the warrant? >> >> Patrick >> >> >>> Or you want to assert a more >>> complex rule set consisting of many interrelated graphs. Having >>> separate >>> warrents in these cases just unnecessarily blows up the metadata. >>> There is >>> also no problem with signing several graphs at once because the >>> SignatureMethod can define how the graph set gets canonialized. >>> >>> >>> >>> So we could define: >>> >>> >>> >>> <rdfs:Class rdf:about="&swp;/Commitment"> >>> <rdfs:label>Commitment</rdfs:label> >>> <rdfs:comment> >>> A relationship between an authority and one or more graphs, in which >>> the >>> authority commits itself in >>> some way to the graphs. Commitments may include a digital signature >>> by >>> the >>> authority. >>> </rdfs:comment> >>> </rdfs:Class> >>> >>> >>> Chris >>> >>> >>> ----- Original Message ----- >>> From: "Patrick Stickler" <patrick.stickler@nokia.com> >>> To: "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com> >>> Cc: "ext Pat Hayes" <phayes@ihmc.us>; "Chris Bizer" <chris@bizer.de> >>> Sent: Wednesday, April 07, 2004 3:25 PM >>> Subject: Re: rewrites for paper sections >>> >>> >>>> >>>> On Apr 07, 2004, at 15:58, ext Jeremy Carroll wrote: >>>> >>>>> >>>>> We should consider whether Warrant is misnamed: possible other >>>>> names: >>>>> Publication >>>>> PublishingEvent >>>> >>>> I don't think that the warrant (or whatever it is) equates to >>>> a publication event. The latter requires more than just the >>>> association of authority, signature, certificate, etc. with >>>> a graph -- i.e. the graph also has to be, er, published. >>>> >>>> It's really a kind of stamp, signette (sp?), brand, etc. of >>>> the graph which can be authenticated, and thereby allow the >>>> graph to be authenticated. I.e. a certificate of authenticity. >>>> >>>> (too bad Certificate is so overused...) >>>> >>>> It's a tool used in publication, not the publication itself. >>>> >>>> But I'm quite open to alternatives to Warrant. >>>> >>>> Can't think of any at the moment though... >>>> >>>> Patrick >>>> >>>>> >>>>> ... ??? >>>>> >>>>> Jeremy >>> >>> >>> >> >> -- >> >> Patrick Stickler >> Nokia, Finland >> patrick.stickler@nokia.com >> >> >> > > > -- Patrick Stickler Nokia, Finland patrick.stickler@nokia.com
Received on Wednesday, 14 April 2004 02:16:04 UTC