Knowledge Networks for Pervasive Services

Technologies to pervasively acquireinformation about the physical and socialworlds – as needed by services to achievecontext-awareness – are becomingincreasingly available. Paradoxically, the riskis to make pervasive services overwhelmed bygrowing amounts of contextual data, andunable to properly exploit them. This calls forspecific approaches to automatically organizeand aggregate such data before delivering itto services. Contextual data items should forma sort of self-organized ecology within whichthey autonomously link and combine with eachother into sorts of “knowledge networks”.This can produce compact and easy-to-bemanagedhigher-level knowledge aboutsituations occurring in the environment, andeventually can make services able to easilyacquire “situation-awareness”. In this paper,after having framed the key concepts andmotivations underlying “situation-awareness”and our “knowledge networks” approach, wepresent the design and implementation of a“knowledge networks” prototype, intended asa tool to support self-organization and selfaggregationof contextual data item and tofacilitate their exploitation by pervasiveservices. A representative case study in thearea of adaptive pervasive advertisement isintroduced to clarify the concepts expressed,to exemplify the actual functioning of thetoolkit and of some specific algorithmsintegrated within it, as well as to evaluate itseffectiveness