In this article, we are presenting a series of ontology-driven mechanisms for system diagnosis. Mainly, we introduce an event ontology and concepts related to semantic tag clouds and show how to manage the activities to build an ontology-based diagnosis. We formalize these concepts in order to derive diagnosis actions and to validate the successful actions. We also propose an augmented action model and an augmented event model, and show how they capture semantics not captured by other models so far. We conclude with a use case formalizing the diagnosis decisions.