Friday, June 16, 2006

Semantic Web - apparently it’s ready to implement ... if we knew what it was!

Flicking through some of the recent copies of free IT magazines that are piling up at home, I read in Computer Weekly that the semantic web was ready to implement.

Pardon? The what internet? Semantic?

Basically, having finally got the internet to the stage that it’s usable by mere mortals without ponytails or degrees in Computer Science, one of the next waves of innovation is to make internet content machine readable.

Is this some Doctor Who Cyberman plot? Machines taking over the world?

No. Today we spend a lot of time in search engines trying to find what we’re looking for. If the information was tagged in a consistent way, and available in a machine-legible format, computer applications could search for the required information, interpreting and using the results.

Imagine software that searched for someone selling a particular book through Amazon or eBay at or below a particular price, and placed the order. A simple mashup I hear you say.

But Amazon and eBay offer very different user interfaces, and their programmatic interfaces are pretty distinct too. And it would be ideal if the software could look at other retailers - without needing specific bonding code to be written for each new one. (After all , if I was to do this search manually, I would be able to suss out the prices of lots of online shops - even without resorting to froogle - without having to research how their shops worked. I’d just scan the prices and find the cheapest.)

Tagging is very Web 2.0. Adding Technorati tags to posts is all the rage - and very useful. But as you’ll know if you have ever tried to tag a film review post, it’s difficult to know which tags to add: movie, movies film, films, review, reviews, cinema?

For the semantic web to be usable, common predefined and public dictionaries of tags will have to be widely employed. It’s called an ontology - a data model that represents a domain, used to reason about the objects in that domain and the relations between them.

So if Amazon, eBay and thousands of other online retailers adopt the same standard of tagging for their store catalogues, software agents will be able to understand what is being offered. And if the purchasing functions are labelled, that might just make the semantic web vision closer to being realised.

And given that Tim Berners-Lee is involved, we’d do well to listen the inventor of the World Wide Web. He has a good track record.

No comments: