Formal Nakedness Apology

Last semantic web meetup took place on Feb 7 at the usual place and time. Martin, Vamshi and I were there. We glossed over the following topics :

• To whom does your life, seen as information, belongs to? The intuitive answer (you) seems not to be the correct one. For instance, what you do and think at work, what you do with your friends, etc., certainly does not strictly belong to you. Actually, it looks like nothing about you strictly belongs to you!!! The corollary to this is that any information (including art, business processes, software, scientific ideas) belongs to no one in particular, thus it belongs to all. This is could be thought as the sociothermodynamic nature of information : no information is ever created, information representations are aggregated from already existing information. It's information theory combined with memetics. It's the economics of folksonomies. That's a little bit what Martin talks about in [FR] La moisson de l'implicite and [FR] Causalité et interactions cognitives d'une idée. He says that with little information about what you are and what you are working on and how you are doing it, your employer could induce new information that would belong to him by default, under the current employment laws. Maybe that's just me, but it now seems like a good idea to start writing an Open Information Disclaimer to your friends, family, employer, business relations, etc.
• How science and any cognitive process can be thought as a formalization/discretization/canonicalization machine that compacts knowledge from the outer, more gratuitous strata of the universal episteme sphere into the more formal, discrete hard core of verifiable and computable propositions that is pure science.
• On the why and the how formal nakedness is forcibly a coercively inclusive process : once one gets naked, those around get automatically naked without them necessarily wanting to. This holds because no one knows what can be inferred from the information offered by one's formal nakedness. The uncontrollable nature of formal nakedness paradoxically is the reason why one should voluntarily get naked : because it gives one first degree involvement on one's own representation, thus somewhat enabling one to have a control on the said representation.

Elsewhere on the web, formal nakedness is making its way : on boingboing, read How computers change writing a post on a writing tool dynamically suggesting related memes as you type a text. Also, you'll probably want to have a look at Lion Kimbro's How to Make a Complete Map of Every Thought you Think project.

The Dataïsland Framework Project and formal nakedness engineering

And on Feb. 17, Vamshi sent an email entitled Community Driven Collective Canonicalization and Annotation Agents to Luc Faucher, Petko Valtchev, Zied Zaier and Robert Godin, the group that will be gathering at my office on March 1rst. Here are some excerpts along with my reply :

[...] I would find most satisfying to start work on would be a slight modification to current blogging technology (do correct me if I'm wrong and this kind of thing already exists) -- A means for storing community specific taxonomies which have records containing

1. Links (Bookmarks)
2. Personal Annotations of the webpages that the bookmarks specific to the community for which the database is being created.

The portions of the your personal annotation that you think is relevant to the community in question (in the extreme case this could be just 2 people having a conversation) is replicated to the personal-database-copies of each person in the community. This should include semantic search capabilities that is driven by the community defined ontology.

Well, as I explained in my formal nakedness post, such a tool is in the making. Basically what you have here is the a Dataïsland-compatible platform for each member of the community. Each member has his own ontology and the community shares one or more ontologies. So you'd do this by having your ontology either dynamically bridged with all the other ontologies from the community (that's the purpose of the Dataïsland Induction subproject) or you could have explicit <owl:sameAs>, <owl:equivalentClass> or <owl:equivalentProperty>

mappings in your ontologies, cues for your platform to understand that a local concept (class, attribute, relation, instance) is semantically compatible with another concept in the remote ontology (either a community ontology or a community member's ontology).

The key problem for me is usability. I don't want to scroll through
some long blog. I want this to be a sync'ed, local copy, that updates
itself on a need-to basis, and has a quick response and highly
usable/findable interface to information that is driven by both the
community ontology and my personal ontology. (I'm not sure if an RSS
interface to a blog can do this? Pardon my ignorance). I want this to
potentially plug in to other tools that I use (eventually, if some
team comes out of these meetings, and we make software, I want these
annotations to integrate into the CVS with the source-code as a design
rationale).

You're describing what I call sensors and conduits, here. For blogs' natural language text, there are plenty of tools from del.icio.us to technorati tags. For formal language content, again Dataïsland Induction is meant for just that and will be doing just that in the very short term.

[...] My googling around so far looking for annotation related technology
has had me stumble upoun the Annotea framework. Haven't had the time
to follow up on it enough, and I haven't found any APIs or libaries or
personal agents that use this ... any leads? [...]

IMHO, Annotea is a very archaic way of doing semantic collaborative work. First it's federated, and federated collaborative architectures stinks! Go West and think P2P, here, young man! ;-) TrackBack/Pingback will do the job much better! You'd probably need to increase referential granularity by supporting document fragment and XPath pointers and ranges, but hey, that's easy enough to say that it's done.

Another great talk at McGill

Vamshi sent me another mail today to say that on a recent post, my link to the federated identity management colloquium is outdated because this page now points to another colloquium. So, bad use of the web for a very interesting event. I already scheduled that on my agenda, and will do everything in my power to attend. This time, Carsten Schürmann, Yale University, will introduce the practical and theoretical foundations underlying the Logosphere digital library and discuss an implementation of an executable transformer between two popular proof development tools, HOL and Nuprl. Is it me or Logosphere sounds a lot like Noosphere?