Technologies to pervasively acquire information about the physical and social worlds – as needed by services to achieve context-awareness – are becoming increasingly available. Paradoxically, the risk is to make pervasive services overwhelmed by growing amounts of contextual data, and unable to properly exploit them. This calls for specific approaches to automatically organize and aggregate such data before delivering it to services. Contextual data items should form a sort of self-organized ecology within which they autonomously link and combine with each other into sorts of “knowledge networks”. This can produce compact and easy-to-be-managed higher-level knowledge about situations occurring in the environment, and eventually can make services able to easily acquire “situation-awareness”. In this paper, after having framed the key concepts and motivations underlying “situation-awareness” and our “knowledge networks” approach, we present the design and implementation of a “knowledge networks” prototype, intended as a tool to support self-organization and self-aggregation of contextual data item to facilitate their exploitation by pervasive services. A representative case study in the area of adaptive pervasive advertisement is introduced to clarify the concepts expressed, to exemplify the actual functioning of the toolkit and of some specific algorithms integrated within it, as well as to evaluate its effectiveness.