Warnings for Disjoint Knowledge Omission in Ontologies

Structured knowledge is the most important component of semantic based information systems required to automate intelligent tasks which are normally performed by expert humans. The knowledge about a domain is typically structured in the form of ontologies. An ontology organizes basic terms and the rules for combining these terms in the form that is machine-interpretable and machine-understandable. Therefore, defining an ontology should be considered an important and serious task with well-defined tools and techniques to verify and validate it. A considerable work is being done to define the processes to develop Ontologies for every phase of its life-cycle [1]. Automated tools to evaluate the contents of an ontology are a great help for large scale ontology developers. We formulated some ontologies to represent real life situations and left certain errors in those ontologies intentionally to see whether these tools would raise the alarm or not. To our surprise, there were some important situations where the alarm should have been raised, but we did not get any warning. In this paper, we report those situations and present algorithms to raise the alarm. We have surveyed the available literature with reference to these situations, and unfortunately could not found satisfactory answer. This paper also summarizes the survey. One of the most difficult situation to detect with automated tools is the incompleteness. This requires the intervention of the domain experts / ontology developers. We discovered that disjoint knowledge and exhaustive knowledge omissions are not being handled by existing tools to raise an alarm for the developers. Algorithms have been proposed to tackle these situations. These algorithms have been executed on a number of case studies. We found them working properly.