First there is a need to find an easy way to manage references of how pieces of content are used together to create new content. Something like the way iLife works on Mac OS X. You are writing a text document in Pages, bring up the media browser, drag-&-drop a photo and the use of it in that particular document is stored globally. That way it is easy to see in what documents (and in essence context) this image is being used.

The second part is the need be able to tag all content with metadata in different forms. I don’t actually think there is a contradiction between managed taxanomies and folksonomies. The must successful way is probably to have them interact. Just as only a part of our content can be manually tagged (for the rest we have to settle for some automatic measure) a managed taxonomy can only be that big. A folksonomy however, does not have that limitation. However, the managed taxonomy can regularly be updated based on developments in the folksonomy. Possibly content can have taxonomy tags, public folksonomy tags and even private folksonomy tags they don’t like to share (like Siderean does I believe)

Thirdly is the need to sometimes find a way to create a discrete and unambiguous classification of words in a text. The reason for that is that it allows for a fixed context for that particular word. Also it makes automatic reasoning with advanced queries possible. If I had a text saying “There is a Volvo, a Saab, a Ford and a Toyota on the parking lot” I can ask how many cars there are on the parking lot without the word “car” being mentioned. Also it means that I can ask how many Swedish cars there are since the ontology that is being used contains that information.

Success in the general audience is of course a good way to make things happen fast but do we need to wait for that when there is a need for these technologies already. Possibly the people needing it don’t yet know they need it, but they definitely do know that they have increasing troubles of handling vast amount of information effectively.

Also, “semantic object-referenced word processing” is probably not needed all the time but it could be an interesting alternative to custom database applications to store objects and there relations like used in most law enforcement agencies.

To me it would be interesting to bring that concept to a wider market for example people doing analysis work were paragraphs or even people, locations and things can be combined or at least described/referenced as objects but still looking like a text in a Word document.

I think the development of semantic technologies and OWL/RDF provides some promise in this area but there is still a lack of easy-to-use commercial applications where the content INSIDE the documents can be referenced and tagged. I would love to see Documentum to provide these technologies in the repository. Then we only need a client–in essence a new kind of Office application–to do the actual tagging/combining/writing. Maybe the easiest way is to leverage OpenOffice for a development project like this.