Standardize the problems before standardizing the solutions

I am currently attending the Adaptive Hypermedia 2008 conference which is being held in Hannover, Germany.
Today i participated in the User Model Integration Workshop where some interesting presentations and discussion took place.

The main theme around all presentations was the argument of centralised vs distributed user models. Which one should we adopt and why? pros and cons of each one were presented as speakers took turns in supporting one or the other.
What i found interesting was the presentation of Peter Brusilovsky who emphasized, as it stands now, how long is the road on achieving interoperability across the entire web. He made it clear that among peer systems, the centralised approach works perfectly, by having a common ontology and keeping universal common models. But he concluded that the big picture requires a decentralised approach, since various models for every user are located in systems with different data models among them.

The workshop concluded with a discussion where some interesting opinions were heard.
The participants agreed that re-usability of ontologies must be preferred instead of every one creating his/her own every time. Altough, we understand, agreeing on ONE universal ontology is far from realistic, up to 3-4 ontologies should be something we could achieve in order to accomplish a desired Lingua Franca state among the user modelling community.
Furthermore, user modelling standards, like LOM, LIP and PAPI, appear to fade away, since developers just don't care any more how they gather user information as long they manage it and keep their users happy while doing it.

Something that i really found interesting is the lack of consideration for scrutability and privacy issues from most of the workshop presenters. Only one student, whose supervisor is Judy Kay, the driving force for scrutability in user modelling, presented a scrutable solution which allows users to inspect and alter the way they are being modelled. I am glad that my PhD is not only focusing on interoperability, but also shows how we can add a 'scrutable touch' on our implementations while keeping in mind privacy of user information as well.

Another point that worths mentioning, is the fact that no-one is thinking outside the educational domain. All presented solutions were for achieving interoperability inside the educational domain only...But how about the social networking domain, e.g. Facebook, MySpace, etc.? How about the e-business domain, e.g. Amazon, eBay, etc.? That is another contribution that i think my PhD is bringing to the community.

Finally, every single one implementation presented was following a common (popular) approach: RDF or OWL to describe ontologies, which i agree, and Web Services for systems to communicate and exchange user information between them, which i don't fully agree. Why use such an 'advance solution' when a simpler one that fits the same purpose is available. The RESTful approach is covering its ground against SOAP-based implementations. Facebook, Google's OpenSocial, Amazon and eBay are all providing RESTful APIs. So why don't we as well?

I think the right questions were being asked today and a common understanding of various existing problems which we are facing has been developed. So why don't we build an ontology to map all these identified problems before every one of us goes away and attempts to solve them in their own diverse way? I mean...where we all agree is when identifying these problems; where we disagree is how we should go about solving these problems. Maybe what we should do, is standardize the problems before wondering why we don't have standardized solutions...