The Semantic Web

Tim “I invented the web” Berners-Lee was in town this week giving a lecture at the Royal Society (just off The Mall in London). The Royal Society broke with their usual policy & announced that no tickets were required – given the amount of potential interest this was a brave move, but in the event they got it perfect with the room filling to capacity just as the clock struck 7.00pm. Starting off with a quick history of the web, Tim quickly moved onto his current area of interest: ‘The Semantic Web’. An active research area, this is concerned with providing information in metadata rich document formats, such that machines can perform some form of semantic analysis of the documents.

For example, consider viewing a notice advertising a lecture. With technology as it is, you might cut & paste details into your calendar (having first found the date) & type the postal code in Street Map to find where it is being held & perhaps fill out a form to register for attendance. Where as you really want to do is say ‘I want to go to this’. In the semantic web vision, you literally just select the ‘I want to attend’ link. Your calendar program would grab the time & date from metadata, another agent would find out if it required registration & if so fill out your details & the location would be sent to your GPS enabled phone which would guide you to the destination.

After several years of everyone saying that ‘XML can do anything’, it turns out they were wrong (but at least heading in the right direction), for it seems that an application of XML known as RDF can in fact do everything! Time for proof by example (well with a little handwaving):

Consider your company has 3 XML document formats, each of which have information on different aspects of a product. Now you want to have all this information in one place / format, so you go along to your XML wizard & ask them to ‘do his magic’ to merge the documents. They’ll go off, talk to the people who wrote each format, think a little, and come up with a fourth format with enough elements to represent all 3 other docs. Now how does this all work in RDF ? So, you’ve got 3 documents in RDF & need a single view of all the data. Simple, you just concatenate the files!

Apparently, the key to RDF is that each type of metadata stored in a document is given a unique URI. So even though the XML document may be a tree structure, by matching nodes with identical URIs, rules engines can form webs across the tree. So provided you can markup elements in each of the 3 documents with common URIs for their common metadata, concatenation really is enough. Neat!

Oh yes, here are the slides