Who can predict where the next Google or eBay will come from?

You have to warm to a research paper that smacks of being cooked up in the bar after a heavy day of profound intellectual thought. And knowing some of the people involved I wouldn't mind betting that there is an element of that in "Towards a Killer App for the Semantic Web" a paper out of Professor Nigel Shadbolt's Intelligence, Agents, Multimedia group in the School of Electronics and Computer Science at Southampton.

One of the topics the researches is the "semantic web," or SW as they call it in the paper, which they presented at a conference in Ireland last year. This is not to be confused with Web 2.0, which is more a marketing ploy than a technology.

The SW is one of those concepts that the experts throw around with abandon, but that puzzle the rest of us. Wikipedia has it as a project that wants to create "a universal medium for information exchange by giving meaning (semantics), in a manner understandable by machines, to the content of documents on the Web". It is all about describing stuff on the web in ways that make it accessible in all sorts of more natural ways than using Google to search. You should be able to ask real questions.

The SW seems to boil down to things like RDF, a set of initials that are so familiar to the researchers that their paper does not explain but that may mean. If Wikipedia has it right, it means Resource Description Framework. You have to bundle them together to get an even harder concept to graft, ontologies.

A key feature of the SW is that the "technology could bring an order and consistent structure to the chaotic Web. So information access would be assisted by semantic metadata."

The Southampton team reckons that the SW could have as big an effect on the web as the original Mosaic browser. That's why they believe that "no-one who wishes seriously to address the problems of knowledge management in the twenty-first century can ignore the SW".

So far no one has done much outside the academic world to push the idea of the semantic web. This leaves the way clear for some nimble footed innovators to get in quick. Be warned, though, that the big players do have teams quietly beavering away.

Trouble is, what will the SW do? After all, no one anticipated the web explosion or the emergence of technologies "to enable users to, for example, transfer funds securely from a credit card to a vendor’s
account, download large files with real time video or audio, or find arbitrary websites on the basis of their content".

The people working on the SW assume that this too will see exponential growth. But what will make it happen?

You can just build it and expect that they will come. But then you could fall flat on your face. First you have to lure in the people who will write the stuff that works on the SW. As the paper puts it "the question arises of how developers and users might be persuaded to come to the SW".

What it needs is one of those "killer apps" that everyone talks about, another Google, eBay, Amazon or Hotmail, to pick those listed in the paper. "Killer apps are very difficult things to monitor. They are hard to describe, yet you know one when you see one."

If you were an investor in IT plays, wouldn't you love to know how to spot a killer app in the early stages? That's what the paper is about.

But don't get too excited. There is no checklist of attributes to look for. Instead, the researchers set out to give some indications "of where we should be looking".

"Killer apps emerge in the intersection between technology, society and business. They are technological in the broad sense of being artificial solutions to perceived problems, or artificial methods to exploit perceived opportunities (which is not to say that they need to have been developed specifically with such problems or opportunities in mind)."

Technology itself isn't enough, they say. "Indeed, a killer app need not be at the cutting edge of technological development at all. The killer app must meet a need, and be usable in some context, such as work or leisure or commerce. It must open up some kind of opportunity to bring together a critical mass of users."

The Southampton group has dissected killer apps in a search for key features. They don't, sadly, come up with an infallible litmus test, "there is no algorithm for creating a killer app. They tend to emerge from simple and inventive ideas; they get much of their transformative power by destroying hitherto reliable income streams for established firms".

"The SW provides a context for killer app development, a context based on the ability to integrate information from a wide variety of sources and interrogate it." One thing that the SW can bring to the part is "personalisation".  This has been "a common thread in the development of killer apps". "Personalisation is often the key to providing the higher service quality than the opposition."

Killer apps, it seems, appear when "there are opportunities to make progress on costs, communities, creativity and personalisation".

There maybe no check list to spot killer apps, but the paper offers some pointers. "The SW is more than likely to thrive in certain restricted domains where information processing is important and expensive." To our mind this could mean financial applications.

They also say that "There is unlikely to be much mileage in simply reproducing the ability to do something that is already possible without the SW." So if someone starts banging on about the SW and its techniques but it looks like something you've seen before, trust your instinct.

You might even want to look at this RDF stuff to see how "tagging" information could deliver benefits. The researchers have a kind word to say about Adobe's use of this in existing products.

What he have in the SW is something that "provides a context for killer app development, a context based on the ability to integrate information from a wide variety of sources and interrogate it".

One consequence of this is that "SW technologies might essentially be expected to enable the retrieval of data in a more efficient way that possible with the current WWW which is often seen as a large chaotic library." But, there is always a but, "it may be that the SW might take off in an original and unpredictable direction".

Bit like the world wide web then. But at least we have the history of killer apps there to go on. Perhaps the Southampton team offers a tool or two for investors and inventors who want to avoid throwing money into the no hopers that littered the first internet boom.