POWDER Working Group Blog

The meeting began with a discussion on the Primer document. Many members took action items to begin to create sections of the document that will exemplify their intended usage of POWDER. Early versions of these sections should be in the first public working draft which is due for publication within the next few weeks. There was then a discussion about the processing of URI components that don't lend themselves to regular expression matching - namely port numbers and IP addresses. WG members are working on this and it is hoped that the number of semantic extensions - i.e. RDF/OWL uses that regular tools can't process - will be kept very low. The group also looked at the wisdom and desirability of making the XSLT that enacts the GRDDL transform 'smart'. Should it only insert owl:intersectionOf and owl:unionOf where it is necessary to do so or remain simple with the result that POWDER documents will often include intersections and unions of single classes? Since the meaning in OWL is unchanged and the Semantic POWDER documents are not intended to be produced or read by humans, it was decided to Keep It Simple - and accept extra verbosity even where it's not required, noting that such extra lines may be removed if so desired. Finally, the issue of a conformance section in the grouping doc was also raised - i.e. the lack of one at present. This may need to be added to satisfy Quality Assurance criteria. The group continues to work hard towards a Last Call announcement. The next versions of the grouping and DR docs to be reviewed at next week's meeting and published shortly thereafter should be very close to, if not the, Last Call versions.

This was the first conference call the group has had for a couple of weeks and a lot was discussed. First of all it was agreed that a POWDER processor MUST understand and process POWDER and MAY do this by processing POWDER-S. Publishing POWDER-S is, of course possible, however, this should be done with caution since its processing requires understanding the semantic extensions (i.e. although it's valid RDF/OWL it doesn't mean as much if you don't implement the extensions mapping strings to URIs etc.) Publishing POWDER-S directly may not therefore have the desired semantics. The group then spent time looking at the Primer which, it is strongly hoped, will be published as a first working draft at the end of this month. There was some discussion following comments made on the public list [1 2] and a new internal draft is in preparation. The group is looking into whether it might be appropriate (and helpful) to include examples of using real world vocabularies (to describe fictitious Web sites), e.g. Dublin Core, Creative Commons, ICRA etc. Those examples are all non-profit - would it be appropriate to give examples using commercial vocabularies such as (WG members) Technosite and Segala?. There was a brief look at the HTTP Link issue which is raging on the TAG mailing list. The group was pleased to welcome Tim Boland of NIST to its membership. Tim brings a wide range of skills, experiences and interests to the group from fields including XML technologies, Quality Assurance and Accessibility. For this meeting, we were also joined by Eric Eric Prud'hommeaux who offered further advice on the issue of IRI canonicalisation. This is clearly a many-faceted topic. IRIs are defined, IRI -> URI is defined, HTML form encoding is defined and so on, but where is it appropriate to use each? It seems that the POWDER approach is likely to end up suggesting that processors make a 'best effort' to canonicalise IRIs within the context in which they operate - and this may lead to false negatives. The group is grateful to Eric for offering to continue supporting this aspect of our work. Finally, an important Resolution was passed, congratulating Andrea Perego and his wife, Barbara, on the birth of their son Alessandro!

The group discussed a number of issues related to its charter, in particular the proposed timeline of Last Call on the grouping and DR docs by end April allowing transition to CR before the summer break. The Primer and Test Suite should be in the public domain, although not as far advanced by end April too. This presents the group with a lot of work to do and we must be mindful of what is practical within that time frame, however, the pieces are coming together.

The group is beginning to plan a face to face meeting in June to go through Last Call comments, probably in Darmstadt hosted by Deutsch Telekom. We will need to avoid a clash with the Mobile Web and Voice Browser meetings that month.

Specific topics: how we will document the rules for transforming POWDER into POWDER-S. Various options have been proposed including some sort of formal rule language or detailed documentation within the XSLT. The group resolved to include a natural language description in the documents and provide both the XSLT and the Test Suite - which should be ample material to allow an independent implementation of a transformation engine.

The general rule for the transformation is:

ref -> rdf:resource,
string -> literal (possibly typed),
elements assumed to be embedded RDF/XML.

That is, if you want something to end up as a rdf:resource, encode it in the XML as a ref and so on. RDF/XML may be embedded within the POWDER Doc.

We also briefly discussed the proposal that all IRI constraints should be expressed in POWDER-S as regular expressions. This will minimise the number of extensions needed to existing RDF/OWL tools. Implementing the full set of IRI constraints would require much more work in such tools.

The POWDER WG held its second outreach event on 18th March, hosted by the GSM Association. The event provided an opportunity for our guests to learn what POWDER is and how it can help to delivery more targeted, personalised content to the benefit of content providers and consumers alike. Carl Taylor from Hutchinson Whampoa gave a positive assessment of the technology but warned about the problems that slow adoption could pose and the task ahead to convince content providers of the benefits of adding metadata, no matter how trusted and useful it may be. Another guest, Scott Rose from Arqiva, talked about the difficulty, cost and therefore resistance to adding metadata to large repositories of video – something that POWDER is designed to streamline by allowing data to be applied to many different resources at once.

As well as a series of short presentations of the various use cases, Paul Walsh of Segala and Charles McCathieNevile of Opera Software both demonstrated POWDER being used in user-centric applications. Putting on events like this not only helps to spread the word about the technology, it’s also very useful for WG members to get first hand feedback from potential users. The group hopes to organise further outreach events later in the year.

The following presentations from the event are available as PDFs:

• Carl Taylor (Hutchinson Whampoa)
• Phil Archer (FOSI, WG chair)
• Alan Chuter, Technosite
• Diana Pentecost, AOL LLC
• Kai Scheppe, Deutsche Telekom
• Kevin Smith, Vodafone Group R & D