The bulk of the meeting was taken up with a discussion of the 'new model' sketched out at the face to face meeting in Boston (see previous blog entry) and the follow up work done
, particularly at NCSR. The group is aware that, at present at least, it does not have the necessary skills to work comfortably with OWL, rule languages and inferences. The model of defining a class of all things that are of a certain type (say, mobileOK) and then a class of things that are on a given Web site (say example.org) and then making the statement that "ACME testing says that the set of resources on example.org is a subset of the resources that are mobileOK' makes sense and can be conveyed in text and images, but rules are much more specialised and harder to work with for the non-expert. Since it is anticipated that it will be non-experts that create and work with DRs, this is a substantial problem.
Machine processability is very important - Description Resources have to be machine processable. The question is how much manual work is acceptable or, at least, how much specifically POWDER processing is acceptable? This will be explored on the mailing list in the coming days and weeks.
The group ended with some housekeeping issues concerning the timing of its weekly calls and the date for its next face to face meeting, provisionally scheduled for mid January in Athens.
Finally there was a brief discussion about the rel="powder" issues. Observers at the face to face meeting suggested that rel="meta" was what we should use and furthermore, that any user agent following such a link should expect to receive back triples about that resource. This is something the group will return to in due course but it looks as if rel="powder" is probably not going to be necessary.