There is a very interesting conversation going on at the FB P2P group page.

An excerpt by Poor Richard: “I am imagining a semantic ontology according to which the key ideas and data of this content could be parsed and tagged to form a distributed database using semantic linked-data structures. This would help transition the collective knowledge base of the research, activist, and social entrepreneurial communities into a machine-readable, semantically linked, searchable form.”

And another one, most releavant here: “Another part of the “research information system” I am visioneering is pattern detection and recognition. An ontology gives us a set of semantically charged patterns. Then what we need is a pattern language with which to parse existing content and match it with our ontology. I am thinking of something like the “regular expressions” used in the old unix text editors or in the Pearl and Awk programming languages I once used.”

Now, I have been thinking a while around such ideas. Ontology (e.g. using OWL) as proposed academically has some drawbacks, as it would imply everybody to share and adopt the same ontology. This looks rather difficult.

Without going too much into details, I have been imagining some very different approach. By my observation, collective intelligence works best when it is decoupled in time and space – distributed. As an example, as a bread job I have been implementing a web site recently. For all the pieces I could not immediately find a solution, I browsed the web and found the information I need – from PHP details, to CSS, to javascript. That is standard practice developing and collective intelligence at best.

Another great example is twitter. At its heart are short messages. Like lego pieces. It’s then up to minds to actually mesh-up tweets, make links, etc.

So I am imagining an approach similar to this instead of some kind of platform for a next step. We already know the power of folksonomies, and there are great tools like Diigo out there. The problem is these are non-linked data with limited capabilities for harnessing.

I am thinking of a tagging folksonomy approach which would enrich or publish data to RDF. Having RDF data we than can re-map all this data semantically. We can tag data with geo-tags, and visualize them on maps. They can be tagged in other forms to visualize differently (e.g. connections between projects, groups, etc.). We can tag videos and pictures to make them searchable, etc. People could come up with countless new ways of visualizing data – where Visual Y would pick up…

Interested in opinions if this makes sense. Thanks for reading.