And now, the panel I’ve been waiting for. Tim O’Reilly takes to the stage with three early examples of real companies betting the farm on semantic technologies. Danny Hillis of Metaweb [listen to the podcast with his Minister of Information, with whom Ian and I had a great dinner last night], Barney Pell of Powerset and Nova Spivack of Radar Networks [who also did a podcast with us, is a member of our Advisory Group, and launched twine last night…]

Tim - “Web 2.0 is all about collective intelligence. The Semantic Web is all about collective intelligence too”.

“There’s something really interesting cooking - platforms for building intelligent applications.”

Before discussion, demo…

Visualising the data in Freebase… “An extreme example of opening up access to data”.

“Because this is an application based on this open database [Freebase], if I find a piece of information that’s missing, then I can add that information”. “All applications that use Freebase are smarter because of that one correction”.

“Freebase is not just about people; it has geographical stuff, media stuff, sports…”

Freebase offers a free api so that people can use this data. Eg open source code on Google Code that allows any text box on any website to look up Freebase for text completion…

Powerset building a natural language search. Reads every document, page by page, and builds a more powerful search. eg “Politicians who were killed by disease” - document may not mention a person as a politician, and may not say ‘disease’; look-ups into (their copy of?) Freebase etc able to work out that ‘Edward Heath’ was a politician, and ‘pneumonia’ was a disease.

Similar example here on Flickr… uploaded by Powerset.

And now Nova is up to show twine. Information is out of control. Information overload. Things are not connected. Collaboration is more complex.

There must be a better way, and there is; twine. For the end-user side of the Semantic Web. Lets you share, organise, and find information.

Now showing twine…

As the user uses twine, it builds a profile of you and your interests; identifying people, places, etc. Created by you, and by your friends and peers.

A twine is a place. Everyone gets their own private twine. You can also make twines for groups, teams, networks.

Semantic graph powers twine; the social graph is people and relationships. The semantic graph is everything.

Create a note. Twine reads it, and recognises people, places, organisations etc in the text… tagging the content with semantic tags automatically.

Twine provides simple browser bookmarklet to add objects into twine as you browse the web. Tries to extract structure from the page… Also capability to ‘twine’ emails.

“Databases don’t have to be in one place” anymore.

When twine sees a url, it mines and crawls the site and infers which things are the most useful to recommend ; “a user-generated bottom-up crawl of the web”

Google is about organising the world’s information. twine is about organising your information.

Everything in twine is shareable and editable. All permissions-based.

Important concept in twine is using tags to search and find things. It’s not just a tag, it’s a semantic tag; linked to concepts.

We analyse properties of the semanti graph to find things that are most interesting for you, based upon your background, your connections, etc.

Peter Rip [Radar investor] doing his due diligence in twine.

Marketing team using twine to leverage the collective knowledge of the group.

Now questions from Tim. “Can we get our hands on this stuff today? What’s the real state of availability of semantic technologies today?”

Danny Hillis - “Don’t necessarily characterise our stuff as Semantic Web. The Semantic Web was a particular case of a way to try to do things a few years ago”

Tim - “semantics ties these things together; there’s meaning in this data. I’d always thought that was one of the things about Web 2.0 too… eg Pagerank - links having an additional level of meaning… Flickr ‘interestingness’, etc. The difference is that in Web 2.0 startups it’s tacit, and hidden and proprietary. What you’re doing is making this structure of meaning explicit, and portable, and usable by other applications; you’re all platform plays.”

Danny - “about connections between things”… leads to“ network effect in value”. Silos don’t make sense.

Tim - you’re all extracting entities. Do you all have to do it yourselves.

Nova - open standards, a la W3C. The semantic web as a concept means a certain set of open standards. There are semantic technologies that don’t necessarily support them. If you want an open network effect you have to support open standards.

Danny - we’re not all reinventing things. We’re not all re-extracting it.

Barney - Freebase is generally about people creating content. Powerset is about machines making implicit meaning already in text explicit.

Danny - Freebase is explicitly designed to be used through other applications. Barney’s demonstration, for example, was another application using Freebase data.

Nova - people can create everything, or machines can create everything. Or there’s something in the middle harnessing the wisdom of machines to the wisdom of crowds. That’s what’s interesting.

Nova - With the semantic web, the data becomes portable and connectable - the web of data. This notion of a database, and this notion of a Platform is changing. The web is the platform, the web is the database.

Danny - I’m saying WE are the Platform…

And it’s a wrap.

Tim - “I think there’s really something here”

Tongue-in-cheek reinterpretation of the new semantic web logo from W3C by Laurian Gridinoc…

Technorati Tags: Semantic Web, Talis, Talis Platform, Talis Platform Advisory Group, Web 2.0, Web 3.0, web2summit