Text mining can be used for several tasks relating both to the extraction of domain-specific knowledge and the management of ontologies. These tasks include the identification of associations between biological entities, the extraction of relationships between biological entities, the alignment of ontologies and the generation of ontologies from text. Most of the methods used in text mining to perform these tasks are based on statistical measures, algorithms from natural language processing , machine learning or information content analysis. We believe that although these methods prove to be effective for a large number of applications, their overall performance remains limited as long a no semantic or ontological layer is added in the generation and analysis of text mining data. An ontological layer will allow to interpret the results of a text mining analysis with respect to formalized ontological background knowledge, and can be used to generate an ontological interpretation of the results of the analysis. In such an ontological interpretation, ontological categories and individuals stand in well-defined ontological relations. The ontological interpretation of text mining results would present several advantages, of which the most important include consistency checks, automated belief revision (ontology curation) and ontologically founded data and information integration. The generation and analysis of an ontological interpretation of text mining results are not straight forward, as it is necessary to deal both with inconsistent and incomplete knowledge. Classical logics will prove to be insufficient for such a task. Therefore, a non-classical, nonmonotonic logic together with non-classical inferences such as abduction and induction is required.