Context summarization and garbage collecting context

Typical ubiquitous computing environments contain a large number of data sources, in the form of sensors and infrastructure elements, emitting a huge amount of contextual data (called context) continuously that need to be processed and stored in some context repository. Usually, this data is for software system's internal use to provide proactive services. Hence, it makes sense not to store this entire huge amount of data but to identify and remove some irrelevant data (garbage collecting context), summarize the left over and only store this summarized and more meaningful data. We believe that such a summarization will result in improved performance in query processing, data retrieval, knowledge reasoning and machine learning. Besides, it will also save the storage space required to store context repository. In this paper, we will present the idea and motivation behind context summarization and garbage collecting context and some possible techniques to achieve this.