An ontology-based disambiguation of terms

One of the traditional applications of ontologies in text mining is the use of hierarchy for disambiguation of terms. For example, it may be not clear from the text if the term 'jaguar' refers to a car or an animal. If this term is defined in an ontology which is used for the disambiguation of the text, then a parent class for the class 'jaguar' should be able to provide an answer. However, a problem is that ontologies are not consistent with each other, and perhaps they never will be. Even OBO ontologies, which are designed to be orthogonal, are actually not orthogonal [1]. One way of solving the problem is to use an initial set of 'trusted' ontologies which are consistent with each other, and a set of rules to point other relevant ontologies. Such an approach can also solve a problem of selection appropriate ontologies to support annotation. Currently, BioPortal contains ~200 ontologies and the number is rapidly growing.