PSI documentation: Questions to be answered

Here are a few of the key questions that the European Parliament has been looking at:
- is it useful to create conventions for the structure of PSIDs? What would those conventions be? What would the "parts" of any structure "reflect" (namespaces, topic classes, etc.)
- should PSIDs be structured in a manner that make then re-useable in non-TM environments?
- dereferencing PSIDs: do we have any guidelines/rules for what a PSID should be de-referencable as? (I think we've already said it should be a UR*L*; is that enough?)
- what should a user (human or machine) expect to find at a de-referenced PS indicator? Predictable metadata types (and values from value-sets)?
- how can (human) language-specific information be handled within PS indicators?

I'll follow up with links to some of our working material, just as soon as I've found the identifiers for them...;-)

Peter