That’s ok with me, I guess.

In terms of research, I’m finally getting somewhere: I’ve written my first “paper” (more like a test run for a real paper, which was reviewed in our academic writing seminar and got completely taken apart by fellow PhD students) and gave a presentation to my research group (which was followed by a very interesting discussion with lots of great ideas and input from everyone, thanks for that!).

I finally got all the Java APIs to work together (yay!) and can run experiments on various ontologies, which is quite exciting and insightful. I’m hoping to get some useful information from the results, which I can then use for my first “real” paper. I’ll let you know how it goes… 🙂

I wish you all a nice and relaxing Easter break!

As I’m trying hard to be a good student, computer scientist and geek, I’ll talk a bit more about my research in this post. Might come in handy to simply forward people to this blog, in case they think I’m talking dada when trying to explain my work (which happens in 9 out of 10 cases.)

A very rough outline is given by my “lay summary” I wrote at the beginning of the year – and it’s already surprisingly far away from what I’m focusing on at the moment:

Spot the error – How to repair faulty ontologies

In critical environments, such as medical applications, the correctness of knowledge we obtain from an information system is crucial – errors and mistakes are clearly unacceptable. But how do we ensure that the system contains exactly the information we need, regardless of its size and complexity?

We call a common basis that defines knowledge and helps us manage information a knowledge base, or “ontology”. We can even infer logical consequences from the facts in an ontology: Say, it states that “Leg is a body part” and “Foot is part of the leg” – this implies “Foot is a body part”. Typical ontologies describing medical or biological terms are very large and highly sophisticated: The size of an ontology can grow quickly, reaching up to hundreds of thousands of definitions!

But the vast amount of complex data can cause errors in the system: We end up with incorrect and unwanted information, such as “A leg has five feet”! Which statements in the system lead to the false conclusion? How can we repair the ontology without producing more errors or removing crucial information?

My research focuses on designing methods and tools to analyse and repair the causes of such unwanted consequences. This makes the errors easier to fix and therefore ensures the quality of a knowledge base. Providing these tools to ontology developers helps simplify and speed up the development process, as well as guarantee that the information obtained from their ontologies is correct and reliable.

Easy peasy isn’t it.