NetKernel News Volume 1 Issue 34

July 2nd 2010

What's new this week?

• Repository: tweak to xml-core, new RDFa library.
• Summer Special training offer.
• Status of NetKernel Enterprise Edition.
• Semantic Technology Report.

Repository Updates

xml-core: A tweak to the classloader export interface to export the Stax implementation classes.

*new* rdf-rdfa: A new companion library to the rdf-jena library. This provides RDFa microformat parsers for HTML and XHTML resources. (See below).

[If you're new, NetKernel has a modular architecture and is able to hot-install packages from software repositories. All system updates and ehancements are available through the "Apposite" package manager tool in the control panel. To check for updates click "Admin"->"Synchronize". On the main packages view select "Apply all updates".

There are many optional libraries available that can be hot-installed with Apposite. Browse the package list and install the set of tools that you need to build your solution.]

Summer Special Training Offer

It's been an incredibly busy few weeks for NetKernel, first there was a very positive reaction to Brian Sletten's half-day workshop on NetKernel and Resource Oriented Computing at Uberconf in Denver. Followed by last week, where we were exhibiting at SemTech 2010 at which several of the other exhibitors use NetKernel as their platform for solutions, and, to cap it all, NetKernel was powering the conference's own Linked Data services.

With all this activity we're pleased to welcome a load of new people to the portal, and this will be the first newsletter you'll have received (last week was the first missed delivery in 9 months due to being at SemTech).

We want to acknowledge the growing interest in learning NetKernel. We also want to come clean - if you're coming to ROC from a traditional coding background then you will almost certainly find it different and perhaps a bit disorienting. After all, the fundamental idea behind Resource Oriented development is to step-away from the APIs and instead to focus on the information resources and the first order information architecture.

Over the years, we've seen that learning and getting productive with NetKernel is dramatically fast-tracked if you are able to watch someone who's at the other end of the learning curve. Simple techniques like using the NK toolset to focus and test the iterative "domain-driven" development process. Learning how to think about the information resource model and to structure and compose a fully operational address space that solves the problem without even touching any implementation code...

We find it takes 2-3 days of instructor-led training for an existing Java developer /architect to convert and be fully productive building scalable ROC solutions on NetKernel. So, to give you a leg up the learning curve, we're offering a Summer Special 33% discount on training.

So while everyone's away at the beach its the perfect time to get a blackbelt in, as Brian Sletten independently terms it, "Software for the 21st Century".

Ping an email titled "NetKernel Summer Special" to services@1060research.com if you want to find out more

NKEE Status

We're on the last lap with regard to NetKernel Enterprise Edition. We have a bunch of updates and enhancements, including completion of the feature set for the NK Protocol library, but we'll hold off shipping these until we cut a full preview build next week.

We have one further task which is to introduce some enhancements to the background metadata infrastructure - these allow an endpoint's state to be requested and updated (think "Resource Oriented MBeans") - and will allow us to provide further system monitoring tools etc.

This is the last critical path item before we declare Enterprise Edition final. So expect one more preview build next week.

Semantic Technology Report

It was good to dip a toe into the Semantic Technology world last week. We made many great connections with the community and refreshed some really old relationships - such as with the guys at Epimorphics, formerly my colleagues at HP Labs, and the team behind the Jena RDF toolkit (which, incidentally, underpins NetKernel's RDF tool set).

So what did we take away? Well from the NK perspective we'd call RDF and "Semantic Technologies" a resource-ecosystem - ie a data model and tool set that is oriented around RDF and the microformat variants. NK is resource oriented, to us its all information resources, so its as easy to create an NK solution with XML, RDF or Business Domain POJO resource representations. One key thing was clear, that there is always the need for a heterogeneous mix in most solutions, whether it be inbound ingestion and analysis or outbound service composition and front-line application and services - which certainly fits with our world view.

There are many people exploring how modeling vocabularies and ontologies can provide business value. With the end goal of loosely coupled dynamically composed queryable data. I guess I'd call this "extrinsic enterprise data" - ie data that is in the environment and not confined to the often fixed pre-determined constraints of an RDBMS. (There are enterprise people for whom this will induce a nervous breakdown!)

There are also, large, often government-centered, projects that are doing large scale extraction of triples and creating very large RDF datasets. By implication, the flexible representational model offers an information structure that can't be achieved using traditional approaches. The investment and commitment would suggest this is paying back with new perspectives on the captured data. I suppose I'd classify this use of RDF as "extrinsic data normalization".

I was focused on the day job, so may have missed it, but I didn't see much evidence of real world solutions in the middle-ground. That is, applying all of that modelling in the hybrid world spanning extrinsic and intrinsic enterprise data.

My guess is that this is not something that people don't want to explore. Rather (and you'd expect this from me wouldn't you) I think that the economics of the software problem at this scale just don't stack up. Trying to explore the composite, dynamic architecture's you'd want to play with in a hybrid solution, are out of bounds to traditional API centred development.

Indeed it was precisely this same economic problem that was vexing me back in 1998/9 when the Jena team were starting up. My belief, then and now, was that unless we step away from the APIs then, although you can often build "something", the cost of maintenance and accommodation of change invariably come to dominate. That's why we started on the research behind NetKernel and ROC, to tap the economic and scaling properties of the Web and bring these to bear inside the software domain.

Enough of the self-justification! We did learn a concrete truth: that actually gathering sources of extrinisc RDF data are the on-ramp to building some interesting solutions. We got the message and have taken it on board. You'll find a new "rdf-rdfa" library in the NKSE multiverse repository. It provides RDFa microformat extraction of RDF metadata embedded in HTML and XHTML resources and produces a JenaModel suitable for pipeline processes with the existing NK RDF tool set.

I also have a nagging concern. A concern I've had for more than 10 years. I worry a great deal that the general use of RDF's graph-centric resource model needs a certain degree of care since it is all too easy to create ontologies and queries that are NP or worse to compute.

As I'm sure you'd expect me to say, I feel the answer is to introduce the Resource Oriented ideas of extrinsic constraints and fine-grained discovery of extrinsic recursive representations that can be cached and composited. In short, thinking and implementing in the ROC domain allows the NP characteristic of unbounded set expansion to be identified, surfaced and avoided. So enabling real world computable solutions.

As John Erickson once so eloquently said in the context of building some very large ST prototypes, "Reality, Reality, F***** Reality". By which I've always thought he meant, thermodynamics and economics will bite you in the ass if you don't take care.

If you're working with NK on ST systems or playing with the new RDFa support, let us know how you get on.

Enjoy the weekend.