POWDER Working Group Blog

POWDER Outreach Event More of What You Want When You Want It This was the third and (probably) final outreach event organised by the POWDER Working Group. Hosted by Yahoo!, who have long been interested in the concepts around machine-readable trustmarks through this working group and the Quatro Project which is part of the EU's Safer Internet Programme, the event had a healthy WG member/guest ratio. As with previous occasions, the event was held under the Chatham House Rule which means that reporting is restricted, however, it is safe to say that our guests included TRUSTe, Mpower Media, the MPAA, Secure Path, AT&T, Cable in the Classroom, the Center for Media Literacy, Comcast and more. The working group's output and ideas were well received with several expressions of support. The discussion, naturally, focussed on the issue of trust. Can a machine trust a machine? As POWDER makes clear, the answer to that is no. Trust is a judgement of one person by another – what POWDER does is to facilitate that human judgement. Several organisations in the room made it clear that they are actively looking at implementing POWDER in one way or another, either as a full-blown service or as a test bed. Various group members discussed their own implementation plans and, whilst no one would suggest that there is anything other than a great deal of work yet to do, future adoption of the protocol seems set for a good start. As for adoption by the big search engines - it's clear that if and when there is sufficient POWDER data of sufficient quality (i.e. without spam), then they will be pleased to use it. Any future effect of POWDER on things like position in search results will emerge rather than be announced.

The group is getting ready for the next outreach event at Yahoo in a week's time so there's a focus on making things work at the moment. The XML Schemas are being installed in the correct locations and it should be possible to use a validating XML parser to validate any POWDER doc before the outreach event. Meanwhile, there's a validator well under way as well as a POWDER processor; and work is beginning on the POWDER to POWDER-S XSLT. It's not clear at this stage what progress has been made with the work in Athens. Work has started in two other group members' (large) companies and so there should be plenty of implementation experience building over the next couple of months. The group discussed comments received from Peter Patel-Schneider. A full response will be composed this week and discussed internally before being sent - the good news is that the comments do not appear to highlight significant or unanswerable problems with the Formal Semantics document. Other comments received were discussed and, again, full responses will be sent soon. The group took one resolution this week. Work on the POWDER Processor highlighted the lack of clarity over whether the displaytext and displayicon could appear more than once in a DR. Since a DR may have any number of descriptor sets, and OWL (POWDER-S ) doesn't restrict cardinality of properties it's clear that multiple strings and icons may be presented to a user agent. Since it is not clear how a given UA would handle this situation, the Description Resources document will be amended to advise DR authors only to provide one string and one image to display.

This was a holiday in the US so the European contingent had a brief call to ease back into the work after the summer break. First off, of course, there were cheers for the Last Call announcement and publication of initial drafts of the Primer and test Suite. Two issues were discussed. Firstly, that the XML element <issuedby> in POWDER will be transformed into the wdrs:issued by property in POWDER-S. Whilst there remains some uncertainty about what this points to (foaf:Agent or dcterms:Agent or wdrs:Agent which would either be a union of or sub class of the other two), the XML schema and POWDER-S vocabulary are unaffected and so can be updated quickly. Secondly, following on from the last call comment and subsequent e-mail discussion last month, for the time being at least we are going to use 'describedby' as our relationship type and propose two MIME types for POWDER and POWDER-S. These will be text/powder+xml and application/powder-s+xml respectively which we believe to be in line with RFC 3023, section 2 of which includes "If an XML document -- that is, the unprocessed, source XML document -- is readable by casual users, text/xml is preferable to application/xml. [..] Application/xml is preferable when the XML MIME entity is unreadable by casual users." These are not formal resolutions of the group but indicate our current thinking. The issue of @rel registration is being discussed in various groups, not least the Semantic Web Coordination Group.