Artificial intelligence systems — for instance, in biomedicine — are often based on knowledge bases, which store expert knowledge in machine readable form. Unfortunately, it is not uncommon that mistakes creep in while knowledge bases are created. Such mistakes can have serious consequences, for example, if the system suggests a wrong medication for a patient. Finding such faults is, however, a very hard task, often just because of the sheer size of the knowledge bases.
For this reason, a team of researchers from the Department of Applied Informatics has developed the “OntoDebug” tool, which is used by people from all over the world. In an article that was recently published in the prestigious “Knowledge-Based Systems” journal, the usefulness of this tool could be scientifically validated by means of several user studies. In particular, it was shown that the effort needed for the discovery of faults in knowledge bases is significantly reduced when OntoDebug is used.
For more information on OntoDebug, please check out here.