On my hunt to find real-life implementors of semantic technologies, preferably ones who have finished projects that impact lots of non-computer people, I’ve now had an extensive interview with a group that fits that bill.

Metatomix Inc does it, in a fairly big and interesting way. They started out in 2001 with the specific purpose of integrating information using RDF. Their biggest success has been with integrated justice, where they write what they call a messaging platform to pull data from some 20 diverse databases and integrate it into a triple store. Then justice workers in Florida, Georgia and a few other states can use some UI to do SPARQL queries against the whole thing, live or close to it, and check if some bad guy they just caught has a warrant in a different jurisdiction. (This all came out of the Jessica Lunsford act and related legislation in 2005).

They’re also doing some work in manufacturing, with global aircraft manufacturer Airbus, and in the financial world with State Street Bank. According to CTO Howard Greenblatt, they didn’t really have much resistance to using the new tools they introduced for ontology building. In fact, Greenblatt told me that the aircraft design techies loved having a more flexible way to describe what they were doing, because by using an ontology they could say things like ‘well, this part is usually for purpose A, but in this case you can also use it for purpose B’  that they couldn’t say in an excel spreadsheet. Now, I haven’t talked to the end users, so I’m taking this all on faith, but the project is really out there and implemented.

An interesting question to me: so how does this work help build Web 3.0 for the world? The short answer is, not directly. The reason is that most of the domain ontologies created are proprietary, either owned by the client or Metatomix. The tools are a good part home grown proprietary apps as well – see below for the stack. I asked Greenblatt how he saw the transition to the Semantic Web, where people share data publicly using open ontologies. His answer was pretty much the same one Duane Degler told me a few weeks ago – it would be a transition from the social networks finding a need for structured data. ‘People will realize that you can do more with semantic information layered on top of what you already have.’

So now my vision for how we get to Web 3.0 is a kind of two pronged approach meeting in the middle – on the one side you’ll build up a kind of hidden library of proprietary apps that are highly structured and use strict logic in the enterprise; and on the other you’ll have social networks using lightweight data structures, loose trust networks and fuzzy logic. Meantime people on all ends of the spectrum will get the hang of using some parts of the Semantic Web, so by the time we meet in the middle it will seem normal.

As promised, here is what I understand is a rough outline of Metatomix’ technology stack:

Business Rules apps go on top (JESS, JSR 94, others)
Semantic layer: Ontologies, Triple store (including Jena from HP Labs)
Event Processing/Data Access (the messaging platform, mainly proprietary)
Configuration Platform (OS)

Then to actually write ontologies, they train the domain experts to use tools – this is where Marwan Sabbouh at Mitre said he encountered a lot of resistance – and according to the Metatomix folks, they trained at least the aircraft techies to use a modified version of the Eclipse IDE, for which they’ve written a plugin for ontology editing. Again, that plugin is proprietary so it may not solve other folks’ problems, but that is the basic approach used.

That’s it for this week. Anyone out there making a human-friendly portal comprehensively listing where semantic technologies are being implemented and which tools are in use? Something like Freshmeat for the software tools and domain ontologies with human-written reviews, but also tied to implementations. There is so much going on so fast sometimes its hard to see the big picture, at least for little fish like this writer. I’d like to sort ontologies by popularity and freshness, count end users, and find out what tools are in the heaviest use by web developers – that sort of thing. If I find it, or make it, I’ll let you know.

Select ratingCancel ratingPoorOkayGoodGreatAwesome

Your rating: None Average: 4 (1 vote)