GOD (General Ontology Discovery) is an unsupervised system to extract semantic relations among domain specific entities and concepts from texts. Operationally, it acts as a search engine returning a set of true predicates regarding the query instead of the usual ranked list of relevant documents. Our approach relies on two basic assumptions: (i) paradigmatic relations can be established only among terms in the same Semantic Domain an (ii) they can be inferred from texts by analyzing the Subject-Verb-Object patterns where two domain specific terms co-occur. A qualitative analysis of the system output shows that GOD provide true, informative and meaningful relations in a very efficient way.