In this thesis we focus on distributed knowledge management (DKM) systems where the knowledge is formalised and ontology-based. Current DKM efforts often consider the general difficult case when the knowledge can be inconsistent. We focus on the particular case when the knowledge in a system is coherent, but at the same time can evolve dynamically. Even if only a small part of an organisation's knowledge can be formalised, it should be consistent across different knowledge activities. Also, a DKM system should be context-sensitive and have semantic autonomy, and we formalise both these notions. These assumptions are desirable when a DKM system is used as an organisational knowledge infrastructure.