The structured web (day 1 of DITA 2006)

Scroll down to the bottom if you're wondering what the post title means. First comes my reaction to the sessions I attended.

The day kicked off with Dave Schell sharing some reuse statistics and case studies from IBM, showing DITA reuse at ca.70-80% in sharing content across similar products, with substantial savings being invested back into documentation improvements, such as more tutorials and sample development. In addition he talked about some projects using DITA in IBM outside of the traditional tech doc realm, including e-learning, and on-demand publishing for business partners of customized configuration guides using generated DITA maps.

Frances Gambino's case study from Information Builder also showed some encouraging trends from DITA and CMS investment: deployment complete and costs mostly recouped within a year, from eg savings in translation; same number of people working same number of hours, but efforts shifted from file management and project management activities to doc design, writing, and working with customers. Result: writing costs went down, writers spent more time on more important work, customer sat went up, tech support calls went down.

Over lunch, Don Day (the DITA TC chair) talked about the DITA TC plans for DITA 1.1, and Robert Anderson from IBM providing an informative example of translation capabilities in DITA using the xml:lang and translate attributes.

Alex Povzner of Siberlogic showed off the capabilities of DITA working with his semantic CMS: the story is definitely exciting here, and makes liberal use of RDF and OWL for expression of semantic relationships. Since the relationships are highly structured, I'm hoping we can improve the DITA map story there as well, and enable a DITA equivalent rendering for those relationships, for the sake of authoring relationships outside the CMS with DITA editors, and for the sake of interchange with other DITA projects that aren't RDF-based.

I finally got to meet France Baril, and her presentation on integrating tutorials with conditional content using XSLT overrides to allow content push instead of content pull was pretty cool stuff, and she had to run through it with me after the session before I got what was going on. I'm hoping some of our planned DITA 1.2 improvements with respect to conref will make this kind of capability more generally available and standardized, without the need for extra coding. In the meantime I continue to admire France's pragmatic attitude and inventiveness, and it's nice to have pressure on the TC to get those features implemented.

This is the first full-on DITA conference I've been to (I wasn't at the DITA Europe 2005 conference in the fall), and it really is quite amazing. Everyone here is interested, focused, excited - there are lots of questions in sessions, and discussions in hallways.  It's also interesting seeing the range of people interested: not all tech docs types by any means, and every kind of company and organization you can imagine is represented.

Finally, I want to say something to justify the title of this post: after talking with Alex Povzner about how DITA and semantic web technologies relate, and continuing the conversation a bit over dinner with Paul Prescod of XMetal, I think that my problem is not with the Semantic Web languages, but that I'm looking for something different: a Structured Web, with more of an information architecture focus than an information retrieval focus. While the Semantic Web adds layers of semantics across content units, it doesn't address semantics within content; and for DITA, the focus is about structured content development, from topics up to maps, as part of one coherent architecture, with consistent support for reuse, metadata, and specialization. This does result in some duplication of intent with Semantic Web technologies, but I think this duplication is necessary, to support the end-to-end picture within DITA's integrated set of capabilities. What we need is a bettter interoperability story: the ability to publish DITA maps into RDF is already there to some degree, but we also need some definitions of what constraints an RDFmap would need in order to allow round-tripping back to DITA, so that we can enable a real exchange across the boundaries of the two standards, without making a writer's job any more complicated than it already is.

Or maybe I'm just too tired to think clearly. Time to sleep.

From DITA 2006,

--Michael Focus Areas: BPEL | DITA | ebXML | IDtrust | OpenDocument | SAML | UBL | UDDI
OASIS sites: OASIS | Cover Pages | | AMQP | CGM Open | eGov | Emergency | IDtrust | LegalXML | Open CSA | OSLC | WS-I