Other points that TBL made in his AAAI keynote (as inferred from news.com):
- Developers should use semantic languages and RDF
- Tagging will improve semantic efficacy on the Web
- The Web is the database, and RDF defines its syntax
We all agree that we want more from the Web. Not just pages, but structured data that facilitates recombination. This is essential to so-called Web 2.0 enterprise. Tim’s support of tagging acknowledges, importantly, I think, the role of people at large in driving semantics onto the Web. Not just catalogers, not just publishers, data-providers, and IT departments. End-users. (Never mind that there are no end-users in a graphical world… we’re mostly chain users.) The benefits of [bottom-up] tagging, and its relationship to top-down systems like ontologies, thesauri, knowledge organization systems, and controlled vocabularies, remains to be elucidated in any long-term way. Certainly the hype and expectation is strong, and evidence of value accumulates in the business plans and performance of myriad Start-Up 2.0s.
We can agree that the Web is a database of sorts. The Database, in fact. It lacks the formalisms of relational databases, but embodies the powerful idiom of linkage, making it both naturally object-oriented and graph-based -- flexible, extensible, robust, self-organizing. Unwieldy, too.
RDF is intended to provide some of the missing formalisms, and thereby make the data more interoperable, recombinant, and hence re-useful. It adds little value in a closed system, and hence has experienced only modest adoption. Presumably its value will be greater in the open system of the Web. Of course, most applications have been designed to work in closed systems, leaving RDF a long-suffering next-year’s technology.
The confusion between RDF Schema and XML schema, overlapping but disjoint schema declaration idioms, went on for years, and did little to bolster RDF’s prospects. I'm afraid the W3C itself has to own-up to responsibility for this.
Has next-year arrived with Web 2.0 and the emergence of Web applications? As interoperability and recombination become the main attractions rather than sideshows, RDF and related enabling technologies may rise in importance, finally getting onto page-one specifications. I think this is Tim's hope and expectation.
Google’s Norvig might point out that injecting/extracting meaning from a proprietary commercial perspective… the normalization layer alluded to in Hitchhiker 650… is likely to happen faster and have longer legs than promulgating encoding standards that require well intentioned and well informed users to deploy. In fact, I think that has been a frequent message. The enthusiastic acceptance of so much of what Google does (and does well), is compelling evidence that there is plenty of semantics on the Web... just not necessarily the W3C variety. Will Google thus own the semantic Web, by virtue of having driven normalization-layer innovation? Thinking….
Image: Rowboats at the Center for Wooden Boats on Lake Union in Seattle, taken during the annual wooden boat show. I'm going to spend my next life in their workshop, leaving only for forays to to Allegro for lattes and Magus for used books.