Fast in the wake of Web 2.0, the next generation of the Internet, the so-called semantic web, is already beginning to appear. Ian Delaney reports on the February 22 Mashup* discussion on the the subject.

The semantic web gets explained in lots of different ways. If you want the long version, then some of the documents at the W3C's site on the subject are likely to be of interest. Here is the short version.

The web today is largely written and contructed for the understanding of human beings. For that reason, machines do a rather ham-fisted job of organising the information it contains. Our best search-engines, for example, are entirely reliant on human judgement in order to find relevant answers for our searches. And they aren't able to link pieces of information together in helpful ways.

A semantic web would start to allow machines to understand the subject of pages, the sort of information they contain and be able to combine that usefully. Ultimately, it's hoped that you'll be able to search the web in a manner akin to a database. If, for example, you want a hotel in Cannes with childminding facilities and a pool, then that is what you will get if you search for it on a semantically enabled web.

It's important to disambiguate semantic from synonyms, or the idea of a thesaurus for the Internet. Google is already able to recognise synonyms: if you search for 'cheap cars', for example, it knows enough about language to also return results about 'inexpensive automobiles'. A semantic search would be able to identify that 'Fiat Punto' falls under the class 'cars' and that it is less expensive than a lot of other cars.

The evening's session was largely devoted to what is available now and in the very near future.

Sam Sethi of Vecosys, chairing the session, demonstrated some of the uses that are currently being made of Microformats. These are small chunks of machine-usable information that can be invisibly embedded into web pages. Typically, these might be used to give a geographical location, contact information, descriptive tags or event details. Sites such as Upcoming.org, Twitter and last.fm use these already and the trend appears to be spreading.

By downloading and installing Firefox extensions such as Tails, Operator or Webcards, your browser can detect and read these and then transfer the information to the relevant programme on your computer (on the basis of a quick review of the three, I would recommend Webcards at the moment). Firefox version 3 is rumoured to be introducing native support for Microformats.

Moving on, the first panellist Mark Birbeck is the CEO of X-Port, who make Xforms Player and introduced Sidewinder. [Xforms were originally the new format for making HTML forms using XML, but its use has expanded to include all sorts of user interfaces]. Sidewinder is a web application framework for wrapping up web applications and making them appear and act like desktop applications, giving the developer full control of the appearance and behaviour of windows. This is achieved by adding metadata to Xforms containing these applications.

Paul Walsh of Segala introduced the idea of Content Labels, a method to add machine-readable content information to internet pages. While the web 1.0 approach to this was metatags, which were independentally added by web developers, and thus subject to abuse, content labels might depend on third-party verification. Sites might, for example, claim to meet certain accessibility standards or be suitable for certain age groups, and the author might pay a small fee to be independently assessed as to whether they do, in fact, meet these criteria. Walsh asserted that it is likely that the W3C will eventually require Content Labels which are proposed as a replacement for the ageing PICS protocol for child safety.

Much of the later debate focused around Content Labels, with a couple of quite salient points. Andrew Hardie from DemSoc suggested that the appeal of adding extra data to web documents is rather narrow, especially since we lack tools to be able to do this with any amount of ease. Andrew Scott from Playtex expressed concern over who and how Content Labels are policed and verified. Others were concerned that this might lead to diminishing access to information on the Internet rather than the increased abilities that might be hoped for from its next incarnation.

The last speaker, Tony Fish, suggested that the semantic web programme is overly directed by sciences and scientists. The real problem with the web, he suggested, was that all of its output is 'flat' and he looks forward to a sensory web which embraces the needs and desires of the many, rather than providing more technology for the few. This view was robustly challenged by Norman Lewis from Orange who protested that achieving a sensory web will 'require more semantics, not less'.