Say a Little Prayer for the Semantic Web

Earlier, I had blogged about SPARQL. The feedback I got was along the lines of asking why I was bothering about that since RDF never really took off. Well, it is true that adoption of RDF is pretty weak. I guess that means the semantic web is not really ready for the main stream yet.

Too bad, really, because the problem that RDF was invented to solve is both interesting and compelling. It is the problem of context sensitive search. Wouldn't it be nice if you could write a search query that differentiates between Huckabee the politician, the store chain, or the movie? Maybe this problem isn't serious enough to bother with the trouble of authoring anything as onerous as an RDF document because google is already doing some special parsing tricks to provide some pre-canned context sensitivity. For example, if you google movie: Huckabee, then you see only movie reviews with the word Huckabee in them. That's not exactly what was wanted but it is closer. You can do similar stuff with define: and stocks: but this is not a complete solution to the problem of context sensitive search. In the world of google these are called search operators.

But perhaps it is still premature to declare RDF as dead. The CEO for the Reuters news agency has talked up RDF in a recent interview. They have even introduced a cool tool for generating RDF from textual content. There's nothing like a cool tool to generate traction in the world of protocol adoption. However, this tool still has a little ways to go before it is going to be of much use. I pointed the tool at another blog entry of mine (Data Access Strategies), just to see what it would do.

What it generated could do little more than identify the following from the article as industry terms; web application server, web application development framework, ruby utilities, django utilities, rapid web application development, business application software development, operational systems, and web application developer frameworks. Not very intelligent, eh? It did correctly identify Martin Fowler as a person but incorrectly identified VB.NET as a company. To give you a basis for comparison, I went to the bother of authoring an RDF of that article myself. That document won't do you much good unless you are comfortable with RDF or have an RDF editor. As I hinted at before, RDF is kind of complicated. Here is a picture of what that RDF looks like when viewed as a resource graph.



Here is a PDF of the graphical version generated by RDF Gravity.

Here's hoping that Reuters continues to work on making this tool better. There is also a recently published extension to the very popular RSS format that allows you to add more context sensitive tags to your site maps. Here's hoping that takes off too.