Thinking Clearly .

Goodness me…it really wasn’t that long ago that I wrote “OWLED 2006 is nigh”. Sigh. Things are always nigh!

This time we’re bigger than ever. We had more submissions (in the 40s over in the 30s) and we have far more registrants (90! 90 people have registered!) than last year. The schedule is quite full, but interesting. I don’t recognize loads of people which is either the good thing that we’ll have lots of new faces or the bad thing that sleep deprivation has finally destroyed my brain. Time will tell…someone.

One of the highest profile and, let’s face it, important parts of an OWLED is the “standardization” bits. In the first OWLED, we agreed on a set of features that became OWL 1.1. At the second, we agreed to submit OWL 1.1 to the W3C and pursue a working group. This OWLED we’ll agree to….whatever we’ll agree to. It’s pretty open.

When we started OWLED, I, at least, wanted some institution that would ensure that there was some hope that the work I and so many others were doing would have a sane route to adoption. I also wanted a good way to deal with the “easy” stuff that we all knew how to do, had heard people clamor for a million times, and perhaps had already done in idiosyncratic ways (e.g., QCRs, user defined datatypes, etc.). Even before a working group started, I think OWL 1.1 has helped matters. You now can use QCRs and expect all the tools to support your syntax. A small, but important, step.

While a lot of stakeholders are clearly coming out to OWLED 2007, many are not, and we only have two days. I’m hoping I can gather some feedback on needs, thoughts, and desires for OWL in advance. I divide possible enhancements into four categories:

1. OWL 1.1 issues (these should go to the issues list); these are mostly for the working group
2. OWL “1.2” issues; these are small, easy things that would be widely useful but didn’t get into OWL 1.1; inverse functional datatype properties are a good example; richer annotations are another
3. OWL 2.0 stuff; radical but still important!; n-ary predicates, non-monotonic features; fuzzy or probabilistic features; macros; things which we may not know exactly how to do or we might not know what we should do
4. OWL-adjuncts; query languages, rules and the like; language features that plug into or layer on top of owl, and typically have a distinct working group

(I omitted APIs, protocols, and surface syntaxes. These are important too, but sorta separate.)

I, personally, will be pushing for SPARQL/OWL (with mixed TBox/ABox queries) and DL-Safe rules (both under number 4). I think we could write specs for these and have implementations in fairly short order, and I think these would be enormously useful to the community. SPARQL/OWL, at least the core that I’m thinking of, is definitely a “OWL 1.2 adjunct”: easy, mostly implemented, users all over; let’s write that spec!

So that’s my brief. What’s yours? Please please please either add a comment to this post, send an email to public-owl-dev, or drop me a private note (preferably with a subject tag [OWL1.2], [OWL2.0], etc.) describing your feature, topic, or use case. I will try to filter through it all and present some sort of coherent…thing…to guide discussion.

You don’t have to be at OWLED to make a suggestion! If you aren’t there, I can’t promise that it will be as well championed as it would be if you were there, but I shall do my best.

Most of the work must, of necessity, take place after OWLED (or between OWLEDs), but there is an enormous amount of energy that comes from sitting in a room with a bunch of enthusiasts and making a joint decision.

We’ve been busy with new internal projects, as well as some milestones for some customer work:

First, we’re getting very close to putting an RDF-powered data integration app into production at NASA HQ. There have been Semantic Web projects at NASA, but not very many have been put into production alongside ordinay business apps. If we’re not the first (and we probably aren’t), we’re at least early.

At the Semantic Technology conference next week our customer, Andy Schain will be giving a talk about this app, which is called BIANCA, and another one, called POPS, we’ve been working on for NASA.

BIANCA provides a single, integrated view of information about, including relationships between, applications, servers, network services, networks, and change items for NASA HQ. The integration is over four different data sources, and with this integrated view we’re able to provide some novel analysis services, including the ability to do disaster and other outage planning scenarios based on building dependency graphs of the relationships between BIANCA nodes. From these graphs we can generate outage repair plans (though these are not yet optimal plans, that’s coming in the next version) as well as productivity estimates per hour of downtime.

BIANCA is a Pylons web app (and RESTful web service) in front of an RDF database. (The next version of BIANCA will include live querying of DNS, SNMP, IDS, and other “network fabric” services. I suspect we’ll do something in Sesame by building a new Sail for some of these data services.)

POPS (People Organizations Projects and Skills) is the other app, which will go into user pilot soon: it’s an expertise locator service for NASA civil servants and contractors (all 80,000 of them), which also integrates disparate data sources (this time 6 or 7 of them) using RDF. (For more details, check out the 2006 XTech talk about POPS.)

The interesting bit about POPS is the client user interface, called JSpace, which started off as our clone of mspace, but has since diverged in some non-trivial ways. JSpace translates user input into RDF queries against a data aggregation accessible via HTTP.

JSpace is an example of what the cool kids are calling these days a linked data browser, though we haven’t yet done a good enough job talking about it publicly, so no one really knows anything about it at all. One project for the summer is to get more demo data sets up on our site so people can play around with the webstart version of JSpace.

Second, our first internship program, which we’re calling 2007 Semantic Summer, is already an unexpected success. Honestly, I didn’t think we’d get a single applicant, since we’re new and the program is even newer. But we’ve gotten about 10 so far, several of them from very strong candidates, mostly people working on a PhD in computer science and pursuing a diss topic in Semantic Web.

We’ve accepted four applicants for the summer, and there’s another who may intern with us in the fall. This is all very exciting: they’ll be working on a range of projects, including new stuff for Pellet, the next version of BIANCA, and some of our internal projects.

Third, the 2007 OWLED (the OWL Experience and Directions) Workshop—which we’re proud to sponsor—is coming up very soon, the first week of June, right after ESWC in Innsbruck. I wish more of us were able to attend, but it’s not a cheap or easy trip from DC and we’re swamped with engaging work. We did get two papers accepted; they’ll be presented by our European R&D, i.e., Bijan.

We’ve very excited to see the registration numbers looking good, as well as a very cool program of talks and papers. If you’re into OWL DL, OWLED is the conference.

Finally, watch this space on or around 7 June: we’ll have a couple of announcements to coincide with OWLED which will be worth hearing, especially if you’re into OWL.

Or not so much?

JavaFX...Silverlight...the more venerable Flash (will Apollo take off?), AJAX (“soon” with HTML5?), and Java applets…

So much hype! But, in point of fact, most of the tech is pretty interesting. (See JavaFX Script and the DLR for just two examples).

Of course, if I were to go into Grumpy Old Bypassed Minority Tech person, I’d whinge on and on about how Squeak blew this space. Though, Squeak is doing ok enough, what with OLPC and Scratch. We definitely ought to have hosted a Python and a Javascript implementation, at the very least. Of course, Peter Fisk is doing some pretty damn interesting work bringing Lisp and Smalltalk and their environments. It is the environments that I miss the most. I’d take a Hypercard like system anyday of the week!

Alas, I’m a bit skeptical that all this means we’ll get radically more usable apps overall. It’s possible, I suppose. I do hope it makes effective experimentation with new presentation and interaction techniques much easier.

Today we were looking for a new code name for an internal project, and I realized that it’s also Monkey Monday and we owe the blogosphere a post about monkeys. Putting these two needs together, we started looking for an interesting monkey to talk about and, perhaps, to find a name to steal.

A night monkey (aka owl monkey) is the only nocturnal monkey, has freaky big eyes (more rods than cones), excellent night and spatial vision:

(Thanks to UW’s National Primate Research Center—damn, I wish these guys needed some SemWeb goodness!)

The genus name is aotus, which we decided makes an excellent SVN repo name.