Semantic Web on Steroids

Pierre Lévy now takes the discussion to another level; I heard him give a keynote at AoIR 2003 and was very impressed (and sufficiently confused at the scale of his project), so I'm looking forward to this.

He says that our next major scientific-cultural project should be a constant and all-inclusive online availability of human cultural heritage in a kind of converged worldwide library and museum system. This, then, should be organised by a universal semantic coordinate system.

We are facing a new informational space, which is in the process of universal interconnection - but we have no universal semantics to address this space; Pierre has been working towards this for the last 15 years. This semantic system should be organised in a hypertextual form, in which each part can serve as the principle for another.

To take the example of the history of language and knowledge: we began with orality, managing individual and tribe consciousness; on top of this builds writing (it doesn't replace, but complement), managing cities and civilisations; on top of this, the alphabet as a specific, streamlined writing system (incidentally, this refers to both letters and numbers), managing money, monotheism, and empires; on top of this, mass media (in a wide sense), managing industrial and open societies.

Finally, then, on top of this builds cyberspace, which involves interconnection (or perhaps interconnectivity - the potential for interconnection, which makes every bit of information potentially part of one universal metadocument), ubiquity, and the autonomous capacity of action for all of these digital signs. For cyberspace, he imagines new symbolic systems and intellectual technologies, an Internet memory (the real-time mapping of global collective intelligence dynamics), enabling the glocal management of knowledge societies.

The obvious criticism of such approaches is one which sees universality as an impossibility - but Pierre calls these criticisms 'post-modern prejudices' and points out that some very universal systems are in fact in place today (e.g. the 24-hour time system, the worldwide currency system). In response, he says: 1. there is one interdependent human species and several corresponding anthropological universals (so unity and diversity are not oppositional terms); 2. we should adopt a universal semantic coordinate system to organise the Internet memory; and there was a 3. which I didn't get down in time.

In analogy to his history of communication, he points out successive linguistic layers in cyberspace evolution: 1. the computer itself; 2. the PC; 3. the Internet; 4. the WWW; 5. the Semantic Web as it's emerging now (and he says that the remaining problems with the SemWeb might be due to the fact that there isn't a universal semantic perspective for it); 6. eventually, the idea he suggests: of a Internet Memory using a 'digitong' (his term) semantic coordinate system/digital information architecture/knowledge economy metalanguage.

Whew. OK, as a next step, then: the basics of meaning are ternary, he says (i.e. consisting of three parts): being (or in linguistics, the signified), sign (signifier), and thing (referent). This enables nine potential linkages between two of the three (and the images he uses here are familiar from AoIR last year) - B<-B (feelings), B<-S (message), B<-T (body); S<-B (society), S<-S (thought), S<-T (truth); T<-B (world); T<-S (time); T<-T (space), you get the idea; each with a specific meaning attached. These are the nine digitong anthropological archetypes, and they translate to various philosophical approaches (e.g. materialism = T<-T).

Further, there are six elements in digitong: knowledge (S), intentions (B), skills (T) on the cognition/virtual side; semiotic operations (S), social roles (B), technical functions (T) on the phenomena/actual side, and there can be a large number of interrelations amongst these (e.g. semiotic operation <- social roles = communication). 6 x 9 = 54 archetypes; 6 x 81 = 486 types. These many variations therefore serve as a coordinate system to identify information semantically.

Phew - this is brave stuff; seems to make more sense somehow when I'm jetlagged. I don't know if it can work - it appeals to a desire to organise the world, but is this a fool's hope of coming up with a grand unified theory of everything? Pierre says he has translated the Learning Object metadata standard into digitong, but I'm not sure if learning objects themselves aren't fundamentally flawed...