When I was a whippersnapper, maybe 12 or so, circa 1960, computers were coming into the public eye. Curiously one of the first real world uses in the UK was in the accounts department of Joe Lyons’ chain of coffee shops, neatly combining two of my future passions—but I digress. A personal interest at the time was politics and I frequented the school debating society and some memorable political meetings. I can’t say that I was terribly successful as a politico, I wasn’t comfortable with the disingenuity, ‘robust’ discourse where frankly, I was generally gotten the better of. It seemed to me at the time that the obvious thing to do with computers was to use them to find the ‘answer’ to these hotly debated political issues, to replace the debating societies and parliament, and in general run the country. Instead of endless debate you should be able to ‘feed’ everything into the computer and come up with ‘the answer.’ I think that at the time I assumed that the computer would come out on my side of the debate—but I digress again...
Even though this idea seems as ridiculous now as it did at the time to my relatives and friends, a lot of subsequent effort in the field of information processing has been devoted to finding ‘the answer,’ to increasingly complex questions—even if it is not (yet) used to run the country. You can trace the evolution of the computer’s identity in how people describe it—as a ‘computer’ for doing stuff with numbers, an ‘information processor,’ or something used in ‘information management,’ to even more clever devices capable of machine to machine interaction and ‘reasoning.’
Which brings me to our attendance at the first World Wide Web Consortium (W3C) Semantic Web in Oil and Gas (SWOG) workshop held last month in Chevron’s Houston location. Oil IT Journal subscribers can read our report from this event on page 6, and the presentations and ‘official*’ report are also available online.
If you don’t know about the semantic web, I suggest that a quick visit to Wikipedia may help understanding what follows.
I have editorialized before about the semantic web—notably after our attendance at an earlier W3C event**.
Very briefly, Tim Berners-Lee’s (TBL) vision was for a ‘next generation’ world wide web of ‘linked data.’ There is a lot of structured information tucked away in vanilla HTML pages—information about people, conferences and webcasts, company profiles and stock tickers. It would be nice to be able to bring all this together and assemble it into larger chunks of useful information. TBL’s idea was to use a simple ‘triple’ modeling construct—the resource description format, RDF, to tag all this useful stuff on folks’ web sites.
Since its introduction around the turn of the millennium, RDF has been something of a flop. On the one hand the calendars and stock quotes, in so far as they have been standardized (not much!), have mostly been done in XML ‘microformats,’ not RDF. On the broader data front, the hegemony of the relational database is unshaken. Data modelers seeking to innovate have mostly gone for XML over RDF—witness our own industry’s WITSML and PRODML efforts.
I think that the semantic web’s failure is partly due to a subtle difference in the kind of data we are talking about and what we mean by a ‘data.’ Is a web factoid like ‘Britney Spears is dating Prince Charles’ (I made that up) data? How about a statement like ‘the fuel consumption of a Ferrari F50 is 75 liters/100 km’ (also made up)? Capturing the first factoid in an RDF statement is pretty trivial, the second is not so easy. If you have an engineering mindset, you’ll see the potential pitfalls of typos in units, unspecified conditions (at what speed?) and the likely requirement of other representations (fuel consumption in miles per gallon.) Right away we are confronted with a data modeling issue. And believe me, RDF may be simple up front, but modeling stuff of minimal complexity gives some impenetrable RDF ‘graphs’ in no time at all.
The SWOG workshop gave examples of both kinds of modeling. Geological constraints like the ‘Comblanchian is a member of the Dogger’ sound quite similar to the Britney Spears relationship. Such text-oriented factoids can be dumped into a triple store for machine ‘reasoning’. Chevron has reported using RDF to gather collections of factoids in a way that mirrors master data management—or as Chevron’s Frank Chum put it at the SWOG, ‘like a souped-up business intelligence system.’ On the other hand, the FIATECH/POSC Caesar ISO 15926 IDS/WIP*** is built around engineering data relationships like the Ferrari example above. This is an exciting extension of RDF-based modeling into the engineering world. The engineers have had to invent some RDF extensions to achieve this, possibly, for the purists, breaking a few eggs in the process.
The ISO 15926 is a flagship project not just for oil and gas but for the semantic web at large. Not the least for the new peer to peer deployment of the iRING (page 1).
One facet of meetings like the SWOG is the facility with which difficult terms are bandied about. My favorite is ‘ontology.’ If I had a penny for everyone who has heard this term used without having a clue what it means I would probably have enough to buy that Ferrari by now.
After the event, I signed up with the ‘ontolog’ online community***. I was not disappointed. Ontology is an Alice in Wonderland concept that can mean whatever you like. From a list of ‘stuff,’ through hierarchies of more ‘stuff’ to a categorization of the whole of human existence!
I found this of great comfort and have now downloaded Protégé and am busy inputting factoids from the modern politico-economic situation. I’m also hard at work on my political upper ontology—although I’m having a little trouble disambiguating ‘liberal’ from its American and European contexts. But I’m nearly there. Next month I hope to report on progress in fixing the world economic crisis and provide a definitive solution to the liberal/conservative schism that has held up progress for so long. Yes we can!
*** www.oilit.com/links/0901_9
**** www.oilit.com/links/0901_10
© Oil IT Journal - all rights reserved.