Context and context management

Understanding and modeling contexts has become a very important part of the recent developments in Computer Science. Especially in Artificial Intelligence and in Ubiquitous Computing contexts play a vital role. However, it is still not very clear, what contexts are and how they are managed. In this thesis, we propose definitions for context and for context management based on a broad examination of literature on context from di erent fields. Based on these definitions, we design a context management system and conceptualize its operational behaviour. The foundation of this context management system is an extension of Bernd Mahr’s Model of Conception (MoC), which serves as the knowledge representation layer of the system. A first mathematical model of the MoC has been created by Tina Wieczorek by utilizing the so called e-set-theory. e-sets are less restricted than their counterparts in Zermelo-Fraenkel set theory with Choice (ZFC) and therefore were the appropriate tool to model the MoC which is highly self-referential. In order to also model the extension of the MoC presented in this thesis, we again use e-set-theory and transfer modeling concepts like pair or function from ZFC to the e-set-theory. As a second step we introduce a theory of modification operations on e-sets in order to understand the consequences that a change of an e-set might lead to. With these additions to e-set-theory, we are able to construct a new mathematical model of the extended MoC. While the MoC serves as the foundation of the context management system, it does not cover all aspects of it. The cognitive system DUAL by Boicho Kokinov describes the missing aspects very well: It provides the means to model and manage dynamically changing contexts through the concept of activation spreading. To finalize the context management system, we transfer and adapt the needed parts from DUAL and thereby come up with a system that can handle contexts as described by our initial definitions.