The Semantic Lunch

Lunch today with John Davies, who’s in charge of next-​​web research for BT. It was quite a long, or rather intense, dis­cus­sion, so I’ll only tackle the basics here. I’ve been trying to nail this semantic web issue for some time, but every time I start reading an academic paper, my atten­tion seems to wander off. So this was a good oppor­tunity for me. I wasn’t going to deviate. As soon as he sat down, I was in with my care­fully prepared, top journalist’s question: “so what’s this semantic web thingy, then?”

It turns out that that is one of the more dif­fi­cult ques­tions. (Damn!) It depends on what you mean. You might mean turning the billions of existing web sites semantic or only about possible future sites or services. The second of these options is the most likely outcome at the present. Semantic web is partly about annot­ating web pages to make them amenable to machines. John prefers the expres­sion ‘semantic tech­no­lo­gies’ to avoid this confusion.

At the moment, inform­a­tion on the web is pretty much designed for human con­sump­tion. You and I know when we go to a shopping site that the figure in bold is the price, that a certain number is the product code and that this piece of inform­a­tion is about the shipping inform­a­tion. To a machine, it may make no sense what­so­ever. If machines are to be able to bring together all these dif­ferent pages to make the web more useful, then they need to be able to read them.

We’ll see the first applic­a­tions of semantic tech­no­lo­gies in the enter­prise space. Its need is more acute. They have lots of data­bases, all built by dif­ferent people according to dif­ferent rules. Integrating the inform­a­tion from those is already a very costly and time-​​consuming activity. One database may talk about CustomerName, another may refer to CustomerID, for example. Joining these things together, so perhaps, a support depart­ment knows about what equip­ment the logistics depart­ment has installed for a customer, improves business effi­ciency. Semantic tech­no­lo­gies put what Davies called a “wrapper” around these dif­ferent data sources to create over­arching access, con­necting dif­ferent data­sources in a way that doesn’t require nearly so much human effort.

People devel­oping semantic tech­no­lo­gies work by devel­oping an ontology for under­standing the sort of data it’s looking at and the tech­no­logy will be able to do some reas­oning based on this. An ontology means a form of clas­si­fic­a­tion system for whatever it is that’s being examined. For foods, that might include their ingredi­ents, nutri­tional prop­er­ties, sup­pliers and type. It’s not just a list, though, but will also under­stand the rela­tion­ships between dif­ferent items. An ontology developed for food might come across E101 and addit­ives and CrispyPop bar. It will know that E101 falls under addit­ives which are part of the ingredi­ents of the bar. If that descrip­tion then gets combined with a database of shops at a whole­saler where you might send the bar, then the semantic agent will cal­cu­late that health food shops aren’t going to stock CrispyPop bars. It’s not intel­li­gence in any way, but the applic­a­tion of rules that the creators have decided upon.

Because the semantic tech­no­lo­gies are light­weight and open source, they are poten­tially avail­able to any company. For this reason, enter­prises that get some of their data from external sources will still be able to use semantic approaches to integ­rate and drill the ensuing com­bin­a­tion. These are my words, not John’s, but one way to think of the tech­no­lo­gies is as providing a toolkit for more easily creating web mashups. Companies already exist, such as Cerebra and Ontoprise, to sell ways to integ­rate enter­prise inform­a­tion using OWL, the web ontology language.

I kind of under­stood, so far, but I needed a good example. I suspect you may be the same.

BT works closely with the National Health Service in the UK. The Service has already gone a long way into digit­ising and col­lating its inform­a­tion through the Electronic Patient Record system and also inform­a­tion on medical know­ledge through the SNOMED clas­si­fic­a­tion system. Unfortunately, though, the data can still be very dis­persed. The X-​​Ray depart­ment might have a patient’s data on a dif­ferent system to the Pharmacy, for example, and those might be com­pletely uncon­nected to the systems used in a dif­ferent hospital or by a clinic.

What semantic web tech­no­lo­gies bring to this is unity and also what John called Description Logics. It can prise open the dif­ferent data­bases, allow an overview, but also cal­cu­late with it. Imagine a patient’s medical record says that they are allergic to almonds. Then a doctor misses this and somehow pre­scribes a nut-​​based food. When the nurse enters this into the patient’s record, the semantic applic­a­tion will use its ontology to work out that almonds are nuts and that there­fore this is a very bad idea. Semantic tech­no­lo­gies that can perform cal­cu­la­tions like this, and poten­tially save lives, are already in use in the UK Health Service.

I’ll leave this there for today. My head hurts already. If there’s interest, I’ll be happy to do a follow-​​up on another day.

Share this post:

Possibly related:

• Threepointouch
• Man About Town
• Another UK Tech Site Launches (Yawn)