In order to tailor web-content to the requirements of a device, it is necessary to access information about the attributes of both the device and the web content Profiles containing such information from heterogeneous sources may use many different terms to represent the same concept (eg Resolution/Screen_Res/Res). This can present problems for applications which try to interpret the semantics of these terms In this thesis, we present an architecture which, when given profiles describing a device and web service, can identify terms that are present in an ontology of recognised terms in the domain of device capabilities and web service requirements The architecture can semi-automatically identify unknown terms by combining the results of several schemamatching applications. The ontology can be expanded based on end-user’s interaction with the semi-automatic matchers and thus over time the application’s ontology will grow to include previously unknown terms.