Information regarding mental illness is dispersed over various resources but even within a specific resource, such as PubMed, it is difficult to link this information, to share it and find specific information when needed. Specific and targeted searches are very difficult with current search engines as they look for the specific string of letters within the text rather than its meaning. In this paper we present Thinking PubMed as a system that results from synergy of ontology and data mining technologies and performs intelligent information searches using the domain ontology. Furthermore, the Thinking PubMed analyzes and links the retrieved information, and extracts hidden patterns and knowledge using data mining algorithms. This is a new generation of information-seeking tool where the ontology and data-mining work in concert to increase the value of the available information.