If you work with pie-in-the-sky technologies like this semantic web stuff you might have read about on this blog, then you need some kind of grounding to keep you connected to reality. You need some kind of compass that points you back to real-world problems. Is this technology useful? What’s missing to make it useful? Why doesn’t it catch on as fast as we like? Where should this technology move to solve people’s problems, to actually save time and money?
For me, this grounding is building small web applications, mostly with PHP. This is interesting and rewarding work. And it’s the lens through which I look at RDF and the semantic web. How could RDF help me here? How could it save me time and money?
One approach that could be interesting is to replace the “M” in the LAMP stack of technologies with an “R”. Instead of feeding the web app out of a MySQL database, it could get its data out of an RDF triplestore. Kendall Clark explores this idea in the context of Ruby on Rails and similar Model-View-Controller frameworks. Basically, it’s about using RDF as the model.
As Kendall points out, RDF is schemaless. You can use it in a /data first/ style, as opposed to MySQL’s /structure first/ style. I think this would be great for agile development and rapid prototyping.
We have lots of technologies in place that could play a role in this RDF web application stack:
- Turtle as a quick-and-dirty way to create some data
- SPARQL to query the datastore
- Some XPath-like mini query language like FSL to pull data into HTML templates
- Object-RDF mappers like Jastor, Kazuki and Sparta as glue between RDF and the business logic
I’ve no idea how all this fits together, but this is something to watch.