All 1 entries tagged Emergent Semantics

No other Warwick Blogs use the tag Emergent Semantics on entries | View entries tagged Emergent Semantics at Technorati | There are no images tagged Emergent Semantics on this blog

September 15, 2004

Emergent semantics (from Deleuze) and the semantic web

Follow-up to Bergson's intuition and reflection in learning from Transversality - Robert O'Toole

Following from that is the semantic theory developed by Deleuze in the Logic of Sense. Meaning works as follows. There is an emerging 'problem' space constituted by the way in which different foldings work against each other. And there is the 'solution' which is a second layer of articulations that repeatedly succeeds in operating on the problem space. This second level is itself a problem space requiring solutions (and so on). The applicability of one problem space onto another as a solution is the site of meaning. The twin articulations make sense of each other. Of course the articulations are not entirely seperate, transversal interactions occur between them, complicated feedback loops shift the relationships between the articluations, and hence change their formations.

This raises a question about the semantic web, which attempts to move us to a model for linking content that works with these emergent semantic processes. The idea seems to be this. A document is the static representation of a web of meaning (problem spaces and solutions interacting with each other). That web of meaning is seen to come from the author. A so called 'intelligent agent' is capable of drawing out and representing these meanings with some kind of schema that links them up with other documents and makes them accessible to others using the same schema.

Such a system was demonstrated in the key note speech this morning. It was admitted that the information structure into which the documents are mapped is no different to a database schema, the advantage lying in the data harvesting tools that can do the job more efficiently than manual approaches. It should be obvious that, considering Deleuze's work on meaning, this is an extremely pale imitation of the dynamic transversal relationship that actually takes place in semantics, in the creation and interpretation of documents. The semantic web could allow the reader of a document to identify a web of significant elements themselves, and then seek connections outside that work with that web. Instead, such systems are prejudging the significance of words within the text, applying a set schema, and linking only to other documents that can be similarly analysed. Nothing creative can come of this.

Perhaps a better approach would be to allow the reader to create their own conceptual map of the significant elements of a document, and then search for other simlar maps that apply to other documents. That really would be a semantic web.