Samuel's Tech Blog

So, one week has passed, or almost one, since my official (degree-)project start this monday.

Will start off with a short presentation of what I'm doing (or going to). My suggested project title is "Integration of DR-PROLOG for semantic reasoning in Bioclipse". Not much explanation there, so I'll cite some from my "approach" section in my project plan:

<cite>
The aim of this project is to integrate an existing tool for reasoning, querying, and integrating different data sources available in semantically annotated formats on the web, and then to compare the tool to other existing technologies (SPARQL and Pellet DL reasoning) in terms of their practical usability as supportive tools in everyday science, or more specifically, in terms of knowledge recovery and performance. In order to accomplish this comparison, real-world use cases such as drug discovery and/or gene expression analysis will be used. As test data available on the web the Bio2RDF (http://bio2rdf.org) database will be used. Finally, visualization features will be added with the Zest visualization toolkit.
</cite>

Someone concluded something like this; "Today there's much work going on to convert and publish data in semantic formats, but not as much on how to deal with that data, connect datasets, reason with it and hopefully gain new knowledge". Most likely, the data - especially when connected - might contain new knowledge to be discovered that cannot easily be found in an automatic way (as of yet), but will need a human mind to discover it. For that purpuse there is a need for good interactive tools for reasoning with semantic data. AFAIS, DR-PROLOG tries to be such a system.

So, a few words about DR-PROLOG then. It's original paper can be found here. "DR" stands for "Defeasible Reasoning" where "Defeasible" has to do with the ability to handle inconsistent and incomplete data in a sensible way. Citing the authors:
<cite>
...defeasible logics deal with potential conflicts (inconsistencies) among knowledge items...
</cite>
The details behind this, as explained in the paper, is something I have to dive into later in the project.

AFAIS, there are two things with DR-PROLOG that makes it interesting for comparison against the Pellet DL reasoner build into Bioclipse already:

1. Being PROLOG-based
   • I've had an keen interest for PROLOG since a long time, and this was in fact the only open sourced PROLOG-based reasoning system for the Semantic Web I could find, and one of only two such systems in total.
2. Using Defeasible Logic
   • This will be interesting to look into as well, to see how reasoners like Pellet are handling the situations towards which Defeasible Logics is aimed.

I have to admit that I'm really quite new to to a lot of Semantic Technologies (having understood the concept from experimenting with Semantic MediaWIki) so feedback is highly appreciated. Feel free to drop a comment below!

The week has so far largely consisted of reading up on different technologies (Java, Eclipse and the Semantic web... *puh* this will take me more hours...) and trying to get a first Hello World Bioclipse Plug-in up running. So far this has involved quite some fighting with Eclipse, but thanks to ongoing work by Egon with the Bioclipse SDK this will hopefully soon be a piece of cake :) Looking forward to that. In the meanwhile I'll continue reading up on stuff.