I was surprised by the content of the Digital Plant event held in Houston late last year (review on page 6 of this issue) and I don’t think I was alone. I was expecting to see and hear from equipment vendors, but the show had been hijacked by the engineering design community—many of whom seem to have been drinking from the ISO 15926 cup. The show’s centerpiece was the iRing demonstrator showing the exchange of plant design data between Aveva, Bentley, IBM and the CIEAM*.
Having just attended the Fiatech meeting in Den Hague, I confess to a little ISO 15926 fatigue—which if you are a regular reader you may share. But this was a great opportunity to check out the progress made on the engineering data exchange front and also on the success or otherwise of the semantic technology under the hood.
In my previous columns, I have been pretty dubious as to the claims made for ISO 15926 and semantics. These stem from my card-carrying ‘old skeptical fart’ status as an observer of previous upstream interoperability initiatives. What appears to be going on in the iRing demonstrator is a file exchange leveraging a standard list of parts. Asking around, I did not get the impression that the data was transferred in RDF—although in a subsequent LinkedIn discussion, I was corrected in that Bentley leverages RDF/OWL in its data exchanges.
Another iRing protagonist put to me that RDF is not particularly germane to the interoperability effort, which is achievable using other formats like XML. But this then means that interoperability is achieved by everyone using the same format. This is what I call ‘vanilla interop.’
The idea behind the semantic web is not that everyone uses the same data format, but rather that data is exposed in such a way that it is ‘discoverable’ and can be ‘linked’ with other data sets. How to publish linked data is explained in a paper from the Web-based Systems Group at the Free University of Berlin.
Data discovery is an interesting idea. It suggests a casual use of a data set by a third party. The approach is used to ‘mash up’ data from different providers—for instance from public data sources like dbPedia and GeoNames. This informal use is rather different from ‘industrial’ usage where more in-depth knowledge of the data sources is usually both available and necessary to achieve what is required. Hence the more rigid iRing approach.
My semantic musing was heightened with the arrival of Honeywell’s ‘Intuition’ on the upstream semantic scene (see this month’s lead). Honeywell, a major automation contractor is taking an interest in the upstream! And in the semantic web!! And in Microsoft’s MURA**!!!
Another interesting contribution to the semantic debate came across my radar recently in the form of a paper titled, ‘A description logic primer.’ Description what? I hear you say. Let me explain. I am indebted to my friend Bertrand du Castel, Schlumberger fellow, initiator of the first online oilfield ontology and author of the oeuvre ‘ Computer Theology’ (CT). I have so far failed to find the strength to read all of CT, let alone write a review, but my take-home from this ambitious homily to things semantic is that the current interest in ontologies stems from a late 20th Century ‘breakthrough’ in something called description logics.
The Description Logic primer*** by three Oxford University researchers looked like a promising starting point to learn more about the breakthrough. The 16 page Primer sets the scene, defining description logics (there are more than one) as ‘a family of knowledge representations used in ontological modeling [...] that provide one of the main underpinnings of the web ontology language, OWL.’ Description logics (DLs) provide ‘formal semantics [that] allows humans and computer systems to exchange ontologies without ambiguity [...and...] makes it possible to infer additional information from the facts stated in an ontology. This important feature distinguishes DLs from other modeling languages such as UML.’
Apart from the fact that the definition assumes a understanding of ‘ontology’ which you may or may not possess, this looks promising and echoes the semantic web ideals of unambiguous information exchange and ‘reasoning.’
The Primer introduces the building blocks of DL ontologies using symbols indicating, for example, ‘inclusion’ as in
Mother ⊑ Parent.
The Primer, using DL ‘SROIQ DL’, shows how DLs handle incomplete information, the ‘open world’ assumption and how adding more axioms like the one above, constrains reasoning. A segue from SROIQ to OWL involves both a slimming down to a ‘lite’ DL and an expansion to a more web-friendly OWL. Those with access to Oracle 11g may like to try the OWL RL reasoning extension to SQL data.
Does DL constitute a breakthrough? The promise of non ambiguous data exchange and reasoning sounds fantastic. Mashups of disparate ‘open data’ sources look promising. Yahoo’s SearchMonkey’s use of multiple vocabularies in dataRSS feeds is a real-world use case, and shows how terminological ambiguity can be minimized. Whether it leads to ‘reasoning’ is another question. What is reasoning anyhow? Is a ‘select’ statement reasoning?
OK, I am now well and truly out of my depth. Fortunately I can see the bottom of the column approaching. You may want to pursue these ideas on the very semantic W3C oil, gas and chemicals business group.
* Center for Engineering Asset Management.
** Microsoft Upstream Reference Architecture. .
*** By Markus Krötzsch et al., January 2012. Cornell University Library arXiv:1201.4089v1 .
© Oil IT Journal - all rights reserved.