The contextual map: detecting and exploiting affinity between contextual information in context-aware mobile environments

Context-aware computing generally focuses on abstracting the situation of individual entities, such as persons, places and objects, making this information available for further computational exploitation. Those resulting entities’ contexts allow a wide spectrum of application cases in various domains, foremost in mobile computing and internet applications. With contexts from multiple entities available, the degree of alikeness of those contexts poses an interesting piece of information. For this purpose, we propose a context model capable of easily identifying affinities among contexts. This multi-dimensional context model is inspired by geographical map models, which are generally applicable for geographical proximity management. We have discovered that geographical proximity can be leveraged to contextual proximity depicting the alikeness of different contexts. Hence, our goal is to apply proximity detection methods from the location-aware computing domain on context-aware computing. The context model has been named the contextual map, representing an entity’s context by a set of multiple contextual attributes in a multidimensional vector. The representation of entities’ contexts as multi-dimensional points in Euclidean space allows the application of location-based proximity detection in order to identify affinities between contexts that encompass far more contextual information than just location. This work presents the concept of the contextual map and discusses its prototypic application in identifying large clusters of similar contexts that aim to facilitate the adaptation mechanisms of context-aware systems. We especially emphasize the utilization of proximity and separation detection on general non-location contexts, hence enabling to dynamically monitor their affinity to each other.