BT plans to squeeze knowledge from the Net

By Alun Williams

Posted on 23 May 2006 at 17:53

With the start of the WWW2006, the 15th international World Wide Web conference, which this year is being held in Edinburgh, BT has been outlining its R&D efforts for the Semantic Web and how to make it more of a commercial proposition.

In a nutshell, the Semantic Web is a World Wide Web Consortium-backed initiative (W3C to give the Web more meaning) that seeks to put more 'meaning' into online data, such that it can be more effectively processed by computers rather having to be interpreted at a basic level by humans. Sir Tim Berners-Lee, inventor of the Web no less, is among those pursuing development of the Semantic Web, at Southampton University.

At the moment, BT's manager of next-generation Web research John Davies told us, computers handle the presentation of information on the Web but people largely carry the burden of interpretation. Search for Rembrandt, for example, on Google and see how you have to sift the Hotel Rembrandt's from the critical text about the painter from links to his works from any other miscellaneous data that may involve the word (the fact that Google's first 10 links are a fairly coherently targeting the work of the artist is a tribute to Google query interpretation rather the raw data itself).

With BT trying to encourage a commercial uptake for the Semantic web, the telco is keen to emphasise what it could mean for businesses, and Davies outlines three main areas of research: advanced integration of information sources, improved knowledge management and development of Semantic Web services.

For advanced integration of diverse data sources Davies argued that the use of pre-defined Ontologies (for 'vertical' business areas, such as auto-manufacturing or dentistry for example), provides opportunities for mapping diverse sets of data together. It enables meta-data to be more meaningfully defined, which in turn means overlapping data sets can be easily joined. The 'ID' of a particular car part, for example, may figure in two sets of data in different forms but the use of associated Resource Description Framework (RDF)-based languages can meaningfully pull the data together.

An example he quoted was BT's work with the NHS in pulling together, from machine-readable sources, information about allergies. If someone is allergic to nuts, for example, Semantic-based processing would have the 'intelligence' to link and understand almond-related data, say, as a form of 'nut'-based information.

Another area ripe for improvement is knowledge management, believes Davies. There is a growing market for tools, working as part of larger systems, that better process and manipulate knowledge from data. Specifically, there is European funding for a SEKT project in this field. Interestingly, this is related to the controversial EU-funded Quiro project (the so-called French alternative to Google). According to Davies the SEKT project concentrates on extracting knowledge from text-based documents whereas the Quiro project apparently has a more multimedia remit.

Finally, there are Semantic Web services. Overlapping with the various WS-* specifications, there is work to be done on providing richer descriptions of Web services that machines can access and automatically process - service discovery and service re-composition (combining multiple services anew).

Davies agree that a key issue is how we progress to the ideal 'Semantic Web of the future' from the massively messy data that currently populates the Web. While those responsible for putting data online will, he believes, pay increasing intention to meta data in order to better identify the meaning of that data, there is also a huge conversion task for retrospectively interpreting - in an automated fashion - existing data.