This chapter takes the complex knowledge systems of metaphors and shows that their structured knowledge can be represented and predicted by ontology. The complex knowledge system of metaphors contains two knowledge systems, source domain and target domain, as well as the knowledge mapping between the two domains. Hence metaphors offer a test case of how structured knowledge can be manipulated in an information system. In terms of the theory of metaphor, we integrate the Conceptual Mapping Model with an ontology-based knowledge representation. We demonstrate that conceptual metaphor analysis can be restricted and eventually, automated. In terms of knowledge processing, we argue that the knowledge structure encoded in ontology, such as the Suggested Upper Merged Ontology (SUMO), is the necessary foundation for manipulating information from multi-domain and multilingual sources. We first extract source domain knowledge structure based on ontology. Next we show that the ontological account allows correct explanation of the parallel yet different use of the same source domain in two different languages. Thirdly, we showed that the restricted set of upper ontology can be combined with the open lexical knowledgebase of wordnets to provide a principled, yet robust, general coverage of language-based knowledge systems.