Although published ontologies are available for many knowledge domains, for complex multidisciplinary areas, such as those covered by many institutions and research networks, they rarely exist. Furthermore, using a general ontology for a limited group of researchers, for instance pertaining to health, would lead to an unfeasibly large set of possible connections. To deal with these issues required a multifaceted approach which included the use of relevant general ontologies, but also deriving specific ones based on a particular community – which in turn needed to combine manual and automatic techniques. For the SPIRES network, a concept map of the expertise of the members of the network was a starting point. At the same time, the general scope of the network – research environments and spaces – and the context of this, which could be considered “research about research” had to be included. A number of information clustering techniques which could generate hierarchical clusters of information were used, together with the ConnectApp tool and other methods.