Cib 2004: thursday¶

Thursday. The day I was scheduled on. There would just be a morning session, leaving me with some time to make a presentation for friday and to take a look at Toronto itself. Could have done that on sunday, but the rain kept me solidly inside the hotel.

In the afternoon I visited the CN tower. High. And I think you can have a lot of fun as a social scientist observing people on the glass floor. The platform at some 400m height partly has a glass floor where you can look straight down to the ground. It's build to carry the weight of seven full-sized rhinos, but still. You need to flip over a mental switch to step on the psychological equivalent of empty air :-)

Doing a presentation invited mr. Murphy to pay me a visit and to put my alarm clock on 07:00. PM that is, instead of AM. So I had a good half-morning of sleep, missing at least two talks that I suspected would turn out pretty good. Bummer. Nico Scholten (TNO, the Netherlands): Building regulations - ICT instrumental to access the relevant clauses

In the Netherlands, 2003 saw a new version of the "building degree" (in Dutch: bouwbesluit), which meant a new version of TNO/BRIS's tool of 1992. A big problem were the long, wide tables you had to use to find the relevant clause. A computer screen just isn't wide enough.

TNO wants to have all standards in XML, then you can finally do some good searching on it. But they aren't built that way. So they do a lot themselves. Putting the laws and regulations into XML with good links between the different clauses is an excellent way to discover mistakes: missing documents, inexisting standards, faulty references... That's the poor quality you get when you create those documents with just a word processor...

Lessons learned from building their new "Bris information warehouse":

• When developing regulations: use such an XML-based system beforehand as you won't find your faulty links in just a word processor.
• Use XML directly.
• People who design think in a different way from the writer of the regulation...
• Standards should have a standard format and standard terminology.
• Cooperation is needed throughout the field of standards and regulations.

Robert Amor (university of Auckland, New Zealand): state of the art in online product libraries

There is a huge variety in information. Most electronic stores are electronic versions of paper-based catalogues. The idea(l): generally available computer-readable catalugue data. Many research projects tried to tackle this.

There are a few commercial systems, almost all of which are classification-driven. A few systems add some parameters to the classification base.

Classification support:

• A wide range of national systems.
• There are EU and UN systems that are almost a standard for a lot of things, as they are the basis for barcodes.
• But there is little compatibility.

Standards sometimes supported:

• IFC and property sets.
• ISO PLIB (he hasn't seen the complex specification implemented in the construction industry, though).

Barriers to development:

• 10000+ products need to be defined, IFC has only 500 classes at the moment.
• National versus international agreements
• Extraction of data from manufacturers (which aren't too snappy to do that).
• It is hard to find a successful business model for online catalogues (RR: VNI/UNETO in the Netherlands seems to have managed for the electrotechnical industry, but mostly for off-the-shelve products).
• it is hard to handle the contextual impacts of project requirements. You get raw catalogue data, whether it fits completely in your specific context...
• Mapping between representations. (He mentioned Kees Woestenenk and the lexicon).
• you need to have support for preferences. (Probably he meant preference for specific suppliers etc.)

Conclusions:

• There is a wide range of efforts
• Major issues that need to be resolved are classes and parameters.

My presentation

It's kinda hard to summarise my own talk. I wasn't too happy with it, but mostly because it's pretty hard to try to get the semantic web across. I really must find some better way to get it across. The idea is good and a lot of people agree that this is the way forward, but... I did find a few comments on the internet last week that should help me get the story clearer.

The main story was that:

• specifications need to reference and access a lot of information stored in other places. And vice versa.
• The semantic web enhances the web's addressing scheme to items inside the data files. So now you can reference (link to) items in an ifc file or a classification just like you can link to other documents on the internet. Ontologies also help dealing with this.
• So the semantic web and specifications fit well together.
• I closed off with some practical examples.