RDF and SPARQL - Just Semantics

Posted under Software by Rob

It’s amazing sometimes the rabbit holes one can find themselves when doing a simple web search.  I started looking for information on the data model used by the census bureau.  I thought it might be interesting to write about.   Instead, I found something more intriguing, at least for today.

Someone had been taking census information, and modeling the data in RDF.  But I’m getting ahead of myself.

Some estimate there are over 30 billion web pages on the internet.  One of the problems with all this information is that most of it is incomprehensible to a computer.  Not in regards to how to present the data, but in regards to what the data means, and how it relates to other data.

This brings us to the idea of a Semantic Web.  This won’t create artificial intelligence, but will allow computers to manipulate the available information for useful ends.  The idea is to relate and use data regardless of what application creates or maintains that data.

Take for example a simple culvert under a road.  One group may install the culvert, another inspects it for safety, another group cleans it, and yet another samples the water flowing through it to check for pollutants.  Each group may use a different software application, even a different database format.  But it would be quite useful to link all these facts to the original real world object.

So along comes the Resource Description Framework or RDF.  RDF is about representing knowledge in a distributed world.  The basic unit in RDF is something called a Triple.  It’s composed of a subject, predicate and object.   You can think of a Triple as a sentence.  The sentence has a thing being described (a web page for example), a specific property (the subject on the page), and the value (which might be tamales.  Mmmm…tamales).

While this seems simple, making it machine readable is the real purpose behind RDF.

RDF can be represented in many ways, with XML being just one of those representations.

And finally, we come to our final new term for the day - SPARQL (pronounced “sparkle”.)   SPARQL  is a quirky kind of SQL used to query RDF information stores.  As long as you’re familiar with SQL of any sort, SPARQL should make prefect sense.

RDF isn’t perfect, and doesn’t solve every problem.  But web searches of the future will need to be more intelligent, and meaningful.  With the proliferation of web sites, massive databases both governmental and private, and everyone under the sun adding content galore, we’ll need better ways to make sense of it all.  Perhaps RDF or it’s progeny will be the answer, perhaps not.

So, while looking for one thing, I found something else far more interesting, and potentially more useful.  Maybe tomorrow I’ll start looking for that Census data model.  Or maybe I’ll look up tamale recipes.  Who can say where the web will take them next.