Samuel's Tech Blog

My project has been more or less on hold for around a week because of exams and other stuff. Looking forward to getting some concrete things done now. While still reading up a bit on OWL, I've started taken the first steps of the prolog/blipkit integration with a simple method for querying a prolog on the form "subject, a predicate, and an object".

I found out that with JPL, you're not really just forwarding a prolog query to prolog. Since Java has a more rigorous way of handling types, one has to deal with types a bit when constructing a Prolog query. That is in contrast to prolog, where determination of types (atom (something like a string constant?), variable, integer or float) is done based on the pattern of the string identifier for that entity. That is, by: Capitalized first letter or not (variable or atom), or being a number with or without a point (float or integer).

The method I created interprets the entities in the Prolog manner and puts together a correctly typed query to be passed to JPL for further passing to Prolog.

If you want to try this yourself, find the plug-in, test prolog file, and bioclipse js code below:

• Plugin: http://github.com/samuell/bioclipse.blipkit
• Prolog file: http://gist.github.com/221649
• Bioclipse js script: http://www.myexperiment.org/workflows/935

Thinking now about what kind of query methods will need to be created later on, but I assume this will become much more obvious as soon as we starts dealing with concrete use cases.