W3C Technology and Society Domain

[participants] [actions] [decisions] [minutes] [irc logs] [venue] [objectives] [agenda] [reading]

1-2nd August 2001 Meeting of the W3C RDFCore WG

Hosted by O'Reilly in Sebastopol

The sections participants, actions, decisions and minutes are draft and have not been approved by the WG

Participants

Regrets

Actions

2001-08-01#1 Mike Dean Provide public URL
2001-08-01#2 Pat Hayes Get Model Theory Introduction Slides online
2001-08-01#3 Pat Hayes Include example in model theory slides
2001-08-02#4 Pat Hayes Re-write model theory as an interpretation of a graph rather than n-triple
2001-08-02#5 Sergey Melnick Send copy of document projected at meeting to the mailing list
2001-08-02#6 Brian McBride Approach Jeremy Carrol to help with RDF/XML language definition.
2001-08-02#7 Dave Beckett With help from Art and Jan, investigate how to represent RDF/XML grammar and the specification of the transformation of RDF/XML to a graph representation.
2001-08-02#8 Dan Brickley Ensure that the document projected, as amended during the meeting goes on line.
2001-08-02#9 Dan Brickley Update RDF Schema spec to reflect decision to adopt conjunctive semantics for domain and range.
2001-08-02#10 Jan Grant Write up a fix/workaround for folks using the currently defined semantics for domain and range constraints.
2001-08-02#11 Graham Klyne Post message to list describing how CC/PP uses datatypes
2001-08-02#12 Jan Grant Post message to list about how EASEL uses datatypes
2001-08-02#13 Pat Hayes Post message explaining why having the typing at the top of the hierarchy in RDFS is not a set theoretical hole.
2001-08-02#14 Ora Lassila Talk to Ian and Frank and get background on why they require cycles in rdfs:subClassOf relationship.
2001-08-02#15 Pat Hayes Discuss cycles in rdfs:subClassOf and rdfs:subPropertyOf with DAML folks and report back to us.
2001-08-02#15 Sergey Melnik Discuss reification alterantives with Pat Hayes.
2001-08-02#17 Dan Brickley Update RDFS prose to reflect decision that subproperties should inherit conjunctively the range+domain of their superproperties.
2001-08-02#18 Jan Grant Propose an explanation of why a subproperty of a transitive property need not be transititive.
2001-08-02#19 Dan Brickley Submit a new working draft of RDF Schema by 6th September 2001.
2001-08-02#20 Ron Daniel Post his list of agreed principles to the list.
2001-08-02#21 Brian McBride Propose structure of documents to be produced
2001-08-02#22 Hyung-Jin Kwon Make his presentation available on the web and circulate URL to the list.
2001-08-02#23 Brian McBride Update schedule to aim for WWW2002 announcement.
2001-08-02#24 Brian McBride Open rdfs-no-cycles-in-subClassOf with Frank Manola as owner
2001-08-02#25 Dan Brickley Update RDF Schema WD to flag possibility of removing no cycles restriction on subClassOf and subPropertyOf and seek feedback if this will cause problems.
2001-08-02#26 Dan Brickley Update schema WD and close rdfs-domain-and-range
2001-08-02#27 Dan Brickley Update schema WD and close rdfs-domain-unconstrained
2001-08-02#28 Dan Brickley Close rdfs-primitive-properties
2001-08-02#29 Dan Brickley Close rdfs-subPropertyOf-semantics
2001-08-02#30 Dan Brickley Clarify rdfs-subClassOf-a-Property as "Are Class and Property disjoint"
2001-08-02#31 Dan Brickley Fix rdfs-online-char-encoding in the next schema WD and close this issue
2001-08-02#32 Dan Brickley Close rdfs-versioning
2001-08-02#32 Dan Brickley Update rdfs-transitive-subSubProperty with explanatory text privided by Jan, include in the next schema WD, but do not close this issue yet.
2001-08-02#33 Dan Brickley Form task force with Graham Klyne, Martyn Horner and Jan Grant to investigate the DAML+OIL approach to XML datatypes.

Decisions

log The WG agreed that the model theory was a good approach to defining a semantics for RDF and would be a useful tool of the WG.
log The WG agreed that nodes in an RDF graph arising from description elements without and rdf:about or an rdf:ID attribute can be distinguished from nodes that had such an attribute.
log The model theory will be defined for RDF graphs, not n-triples.
log Multiple domain and range constraints are permissable and will have conjunctive semantics and this issue is now closed.
log Domain and range constraints on domain will be included in the next version of the schema document and this issue is now closed.
log The issue rdfs-primitive-properties is not a problem and will be closed.
log subProperties inherit conjunctively the domain and range of their superproperties
log The WG will not address the rdfs-versioning issue
log Subproperties of transitive properties are not necessarily transitive

Minutes

The WG thanks O'Reilly and Rael Dornfest for hosting the meeting.

RDF t-shirts provided by Hewlett Packard were distributed.

The meeting objectives and agenda were reviewed and approved.

Establishing a context - users of RDF

Five speakers described uses of RDF, users of RDF and problems with RDF.

RSS

Rael Dornfest described RSS (log) a site description and syndication format based on RDF. RSS is widely used; RSS 0.91 (which is not RDF) is the most widely used form, but its growth is slowing. Users of RSS include:

Rael identified the following issues with RDF:

Prism

Ron Daniels presented PRISM (log),a metadata standard for the magazine publishing industry. The project was initiated to fill a customer need. Its goal is to help publishers deal with customers to do better search, personalisation, alerters, portals and intranets. PRISM users have the following isses with RDF:

It was noted that RSS can be easily used to syndicate PRISM data, which was an unexpected benefit.

DAML

Mike Dean presented DAML as an application of RDF. He referred to the communication the WG has received from the DAML 'joint committee' and went on to highlight the following issues as the most important from a DAML perspective:

XAP

Arno Goudol of Adobe, who was present as an observer, presented (log) information about XAP, an Adobe RDF metadata toolkit which will be launched at Seybold. XAP is already shipping in a limited form in Acrobat 5. Arno's top concerns with RDF were:

Dublin Core

Dan Brickley presented (log) the Dublin Core architecture effort and listed their top concerns:

Presentation of a Model Theory for RDF by Pat Hayes

log

(link to follow - based on draft)

An example was worked through on the whiteboard.

The following issues arose during the discussion:

There was some discussion regarding the role of the model theory in the WG activities. It was agreed that the model theory would be a useful tool of the WG. There was concern that it would not be useful or accessible to part of the target audience for the WG's specifications.

Model Theory and Abstract Syntax Issues

log

The WG considered the following issues for discussion:

It decided to discuss literal is XML structure, the syntactic representation of documents and anon resources first.

#rdfms-literal-is-xml-structure

log

It was suggested that there had been basic agreement on the mailing list, that literals with a parseType of "Literal" would be treated as strings, but the model would contain further information. There are many details to work out, but this is the basic principle. A list of components that would have to be in the model was started:

A number of different approaches were identified:

It was suggested that parseType="Literal" could be dropped but Ron Daniel and Eric Miller spoke up that there are users who use it. It was suggested that the parser adding namespace definitions to the literal might break an XML signature. Concern was raised over entities in the literal. After discussion it was concluded that we needed to investigate the applicability of the XML fragments spec. Unfortunately, no one was willing to own the action.

Syntactic Representation of Document

The question was raised whether there was a need to include in n-triples brackets to bound the statements in a set of statements to mark the scope of the names of anonymous nodes. This was objected to on the grounds that it would break current code which processes n-triples. There was considerable discussion and it was decided not to add brackets.

Anonymous Nodes

(log)

A long and complex discussion of anonymous nodes began with two questions:

On the first question, the objection to being able to distinguish them was that they served no purpose. For RDF which is only making assertions they have no use.

Reasons given in support of them included:

It was agreedthat anonymous nodes in the model can be distinguished from nodes with supplied URI's.

The WG then turned to discuss the second question (log). Amongst a long discussion the following points/issues/observations were made:

In response to a request for use cases where the difference mattered, the following were suggested:

The meeting adjourned to the bar at the end of the first day without reaching any further conclusions and resumed (log) the next day with a proposal that many of these issues were removed if the model theory were defined for a graph rather than for n-triples. The meaning of RDF/XML could be defined in terms of a translation to a graph.

Sergey presented a document analysing the different options in terms of their behaviour from an implementation perspective:

It was agreedthat Pat would update the model theory based on the graph instead of n-triples.

RDF/XML Syntax

Dave Beckett gave a presentationon some experitmental work he has been doing on defining the RDF/XML grammar in terms of infoset. Two questions were suggested:

The following technologies were mentioned:

It was observed that it would be polite to give serious consideration to using tools from the W3C family of specifications.

Jeremy Carroll's new RDF parser ARP was noted in that it is based on a BNF representation of the grammar where the symbols are SAX events rather than characters.

It was decided that Dave Beckett, Jan Grant and Art Barstow would consider consider this problem and come back with a recommendation. Brian was asked to invite Jeremy Carroll to help.

There was some discussion of n-triples and the way its role seemed to be growing. There was some discussion of it causing confusion and being used for machine to machine communication of RDF versus the need to emphasise the point that there can be more than one serialization language for RDF graphs. It was emphasised that the key concept is the graph and RDF/XML and n-triples are serializations of it.

Schema Issues

(log)

rdfs-domain-and-range

The working group decided that mulitple domain and range constraints would be allowed and have conjunctive semantics and that this issue be closed.

rdfs-domain-unconstrained

The working group decided that due to an editorial oversight, the domain and range constraints on the domain property had been omitted from the RDF/XML representation of schema. It as agreed that they should be added in the next draft and that this issue be closed.

rdfs-constraining-containers

The working group decided to make no change to current draft and to close this issue on the grounds that:

rdfs-xml-schema-datatypes

The working group decided to form a task force to investigate the DAML+OIL approach for using XML datatypes. Dan Brickley, Graham Klyne, Martin Horner and Jan Grant volunteered for the task force. The next working draft will explicitly seek implementor feedback on this. We do not consider this a requirement for the next working draft.

rdfs-primitive-properties

The working group decided that it did not believe this was a problem and will be closed with reasons and explanatory text. It is not critical path for the next working draft.

rdfs-no-cycles-in-subClassOf

The issue requests that the restriction on no cycles in subClassOf relationships be dropped. The following points were made:

The working group decided to open this as an issue with Frank as its owner and to explicitly seek feedback via the next working draft and from the DAML+OIL folks.

rdfs-subPropertyOf-semantics

It was decided that subProperties inherit conjunctively the domain and range of their superproperties and the text of the working draft will be modified to make this clear.

rdfs-subClassOf-a-Property

After some discussion it was decided to clarify this issue as "Are Class and Property disjoint" and to do nothing for the next working draft.

rdfs-online-char-encoding

This was an editorial oversight and the editor is actioned to fix this in the next working draft and close the issue.

rdfs-versioning

The WG decided to close this issue without action since it is a known problem that is very hard to solve and is outside the scope of this WG.

rdfs-transitive-subSubProperty

The WG decided that subProperty's of transititive properties need not be transititive and Jan agreed to produce an example to demonstrate this.

rdfs-clarify-subClass-and-instance

The WG decided to take no action for the next WD.

rdfs-isDefinedBy-semantics

The WG decided to take no action for the next WD.

rdfs-editorial

The WG decided to take no action for the next WD.

Next Steps

Dan Brickley will produce a new working draft of RDF schema for 6th September 2001.

Pat Hayes will produce a new model theory by 11th August 2001 and we will use Sergey's proposal for evaluating its impact on implementations.

The WG envisages the following deliverables:

A straw poll of the WG members present indicated that 5 would be willing to work on the primer, 8 on the model, 4 on the syntax, 4 on schema and 2 on test cases. [Aaron volunteered from irc to work on the primer so consider that +2] It was noted that we can arrange for extra telephone bridges to enable faster progress in subgroups.

It was noted that the original WG tried to separate the model and syntax and found it too hard. There has been considerable support for separating these two and the working group decided to try again to separate them.

It was decided that Brian would produce an overall an initial structure for the documents and then we could start writing pieces.

It was agreed that we would like to aim to announce recommendation status at WWW2002.

Short Presentations

Kwon presented the work in which he is involved in the Korean National Computerization Agency developing a national metadata repository. He is looking for advice on whether to use RDF and what tools are available. It was noted that government projects are often required to use standards. RDF Schema having not completed its standards process is an inhibitor to acceptance of RDF. It was noted with awe that we have a whole country waiting for us to finish. Some of our users and potential users are concerned about us getting involved in AI and knowledge representation when what they want is relatively straight forward mechanisms for representing metadata.

Ron Daniels showed a demo done for a customer by InterWoven where the ability to merge metadata from different sources enabled rapid development of an excellent demo.

Eric Miller repeated part of his talk from the semantic web workshop describing some of the work being done in the Semantic Web Advanced Development group using DAML+OIL to develop tools for to support W3C processes. It was noted that the biggest bang for the buck comes from the simpler stuff in DAML.

Close

The meeting closed with the chair thanking the participants, the scribes our hosts O'Reilly.

IRC Logs

Day 1 (logger time)

Day 2 (logger time)

Venue

Our local host is Rael Dornfest (rael@oreilly.com) of O'Reilly (800 998 9938) who has provided us with some local information.

Rael recommends the Sebastopol Inn (6751 Sebastopol Avenue, Sebastopol, CA 95472, +1 707 829-2500) which is across the street from the O'Reilly office. An alternative hotel is the Holiday Inn (101 Gravenstein Hwy So./Cooper Rd. Sebastopol, CA 95472, 707-829-6677) which is 2 miles from O'Reilly office.

The following directions are taken from Direction's to the O'Reilly Sebastopol Office.

From San Francisco:

  1. Take US-101 North out of the city.
  2. Continue along US-101 (approx. 55 miles) until the Hwy 116 West exit towards Sebastopol. (Don't be fooled by the earlier Hwy 116 East exit.)
  3. Left at the light (at the bottom of the exit ramp).
  4. Continue to follow 116 for approx. 8 miles.
  5. At the the first light in Sebastopol (Hwy 116 and Hwy 12) turn right.
  6. Turn left at the next light (Morris St).
  7. Turn into the first driveway on the right.
  8. Welcome to O'Reilly & Associates.

From the East Bay:

  1. Take I-80 East/I-580 West.
  2. Near University Ave in Berkeley, get in the right lane.
  3. Take the exit ramp towards the Richmond/San Rafael Bridge. (Toll $2.)
  4. Take 101-N towards San Rafael/Santa Rosa.
  5. See the SF directions 2-8.

From the South Bay:

  1. Take US-101 North.
  2. Follow US-101 through the city.
    OR
  3. Take 280-North (it will become 19th Ave).
  4. Follow 19th through the city until it becomes US-101 North.
  5. See the SF directions 2-8.

Meeting Objectives

The meeting has the following objectives:

Agenda (Tentative and subject to change)

Tuesday, July 31st 7pm, Garden Courtyard, Sebastopol Inn

Informal get together. Head off for dinner about 8pm.

Wed, August 1st (all day)

         1000 Welcome, Introductions, meeting objectives, agenda review

              Establishing a context - users of RDF
         1030 - RSS         - Rael Dornfest
         1045 - PRISM       - Ron Daniel (unconfirmed)
         1100 - DAML+OIL    - Mike Dean
         1115 - Dublin Core - Dan Brickley        

         1130 break

         1145 Presentation of the model theory - Pat Hayes

         1300 Lunch

         1400 Model Issues:
              rdf-formal-semantics 
              rdfms-identity-anon-resources
              rdfms-graph 
              rdfms-uri-substructure 
              rdfms-literal-is-xml-structure

         1530 break

         1545 Short Presentations - 10 mins each on interesting stuff we are
              doing

         1615 Model Issues continued

         1700 break

         1800 close

         1930 Dinner

Thurs, August 2st (all day)

         0900 RDF/XML Syntax - Dave Beckett
              rdfms-formal-grammar
         1000 short presentations

         1030 break

         1045 RDF Schema - Dan Brickley
              rdfs-domain-and-range 
              rdfs-domain-unconstrained

         1230 Lunch

         1330 Future Plans - Deliverables, roadmap

         1500 break

         1515 Open Mike

         1600 close

Reading

Required

Recommended

Suggested (Late Addition)

Recommended Materials

It is recommended that participants have access to the paragraph numbered Model and Syntax and the RDF Schema specs.


Eric Miller <em@w3.org>, (W3C) Semantic Web Activity Lead

$Id: Overview.html,v 1.29 2001/07/27 16:00:21 bmcbride Exp $