Context Selection and Reasoning in Ubiquitous Computing

Abstract

Context reasoning refers to the process of giving high-level context deduction from a set of low-level contexts. It plays an indispensable role in ubiquitous computing. Most existing reasoning methods are proposed with the assumption that the knowledge of low-level context which is relevant to the given high- level context reasoning is available. When this information is lack, the existing methods blindly select some possible low-level contexts for reasoning, so that useless context might be included. These useless contexts have no or little favorable effect for reasoning and increase computation burden as well as repository burden. To deal with this problem, we use information gain-based method for context selection in our work. Only selected contexts are used for reasoning. Experimental results show that our proposed approach is promising.