I agree that there wasn’t a transition from Web 1.0 to Web 2.0, but there were requirements/suggestions which made a site a Web 2.0 site. But this is all site level, on a more global scale we aren’t, and will never be, entirely Web 2.0 it just doesn’t fit some sites… and nor does DataPortability. e.g. a Bank web application would be silly if it was “Web 2.0″ or “DataPortable”. Web 2.0 on a Global Level is just a fuzzy buzzword.

I think the same will be for “Web 3G” or “Web 3.0“, there will be some fuzzy requirements/suggestions to make a site “Web 3G”. But on a Global Level we will never reach a “Web 3G” or “Web 3.0″ as it is just a fuzzy buzzword.

If we split what you propose for “Web 3G” into smaller chunks, for example:

MicroWeb of Objects (i.e. internal links)
Linked Open Data (i.e. external links)
The portability of data, and the importance of the identity of data
Agents capable of answering Semantic Queries
Automatic Organisation/Classification of Data

These things will be done separately for different use cases, they are achievable. I don’t think there is an explicit need for a “Web 3G” or “Web 3.0″ classification… because the further we advance technology the broader its going to become, and it becomes hard to encapsulate all these things into one. However, there is no avoiding it, there will be a “Web 3.0″ buzzword, which will try to encapsulate these things.

You are indeed correct that the World-Wide-Web was not created to have a Web of Documents. BUT, the documents will never go away as humans will always prefer to read things in a document style. It’s now about exposing the data behind the documents in order to make things easier for machines to produce good results for humans to read (once again in document form).

You are right, there is a distinction between AI and the tools produced by AI research. For instance, inference is distinct from Prolog… but both are used together.

I think you find that Knowledge Bases are taught and historically developed in the Artificial Intelligence / Intelligent Systems field. In my undergraduate degree the Knowledge Based Systems and Knowledge Engineering modules were only taught to the Intelligent Systems field and no other, I think this is the same elsewhere. Also you’ll find that Knowledge Bases are typically a topic at Artificial Intelligence academic conferences. So thats my argument for Knowledge Bases as an A.I. topic

I’ll also argue that Latent Semantic Indexing is an Artificial Intelligence subject as it involves natural language processing (which is entirely an A.I. subject), clustering and in some cases Bayesian logic.

Here lies Web Science.

(p.s. you’ll notice that I have extended this comment into the Linked Data Graph by linking various keywords into DBpedia. This is something which I have got into a routine of doing on my blog. Making this comment queryable)

I don’t believe there will be a big bang. Can you point to the big bang that was the transition from Web 1.0 to Web 2.0? I think it was an evolution and by the time everyone had stopped arguing about what it was it was here. The same is happening right now but people aren’t suddenly going to stop using their existing apps and services. They’ll adopt new ones and cajole their existing ones to catch up once they see the benefits (and those benefits must be tangible).

Also remember that linked data and the web of data are a means to an end, not the ultimate goal. The WWW wasn’t created to have a web of documents, it was created to enable people to access information more easily. The web of data enables the next generation of information rich applications.

For your last point, there’s a distinction between AI and the tools produced by AI research. I don’t think a knowledge base is AI, not is LSI – they’re just techniques for organising information. We need those but they’re not artificial intelligence.

“Web 3G” isn’t an evolution of “Web 2.0″, surely there is some kind of bang into “Web 3G” existence when an app fulfills all of the requirements that make it a “Web 3G” app? (meaning that its not an evolution at all)

The paragraph that goes (I will call this paragraph X):
“A whole new generation……world-wide network of links with meaning”
This is partly really obvious, and partly something that I have been reiterating over and over in my blogs and talks (shame on you for not referencing me ). See here for one such example: http://vanirsystems.com/danielsblog/2008/02/14/my-view-of-dataportability/

Paragraph X is actually the most important paragraph in the whole of your post, because this is actually the original Semantic Web Vision, plain and simple! And Tim Berners-Lee reiterated that in his GGG blog post. We just call it “Linked Data” or “Linking Open Data”, because with RDF – people haven’t crossed knowledge domains with good Semantic URI linking and the “Linking Open Data” project promotes that.

You also forgot to mention that OpenLink Data Spaces (ODS) is also similar to Twine and Talis Engage. I can only speak for ODS but I imagine that Twine and Engage are similar. They are Semantic Web enhanced Social Networking Tools where the user has their own space where they own their data, and all the data in these systems is available through true data portability using Linked Data.

Twine, Engage and ODS enhance the Giant Global Graph or Web of Linked Data… they aren’t the be-all and end-all. Its important to remember that.

Oh, also this statement:
“It is not Artificial Intelligence”
Erm… Knowledge Bases are generally considered a subdivision of Artificial Intelligence.
and this:
“but some of the tools and techniques of AI are needed: neural networks, classifiers, heuristics, Bayesian networks and statistical analysis”
These things have existed on the web for years either as part of search engines or product recommendations. e.g. Latent Semantic Indexing and Self Organising Maps are considered Artificial Intelligence on the Web.

[note]I speak for myself, Daniel Lewis, and not for my employer[/note]