This paper describes an approach to representing, organising, and accessing conceptual content of biomedical texts using a formal ontology. The ontology is based on UMLS resources supplemented with domain ontologies developed in the project. The approach introduces the notion of 'generative ontologies', i.e., ontologies providing increasingly specialised concepts reflecting the phrase structure of natural language. Furthermore, we propose a novel so called ontological semantics which maps noun phrases from texts and queries into nodes in the generative ontology. This enables an advanced form of data mining of texts identifying paraphrases and concept relations and measuring distances between key concepts in texts. Thus, the project is distinct in its attempt to provide a formal underpinning of conceptual similarity or relatedness of meaning.