Using context awareness to improve quality of information retrieval in pervasive computing

Abstract

Publish-subscribe-query information broker middleware offers great promise to users of pervasive computing systems requiring access to information. However, users of publish-subscribe-query information broker middleware face a challenge in requesting information. The decoupling of publishers and consumers of information means that a user requesting information is frequently not aware of what is available, where it comes from, and when it becomes available. Too specific a request might return no results, while too broad a request might overwhelm the user with a combination of useless and buried useful information. This paper investigates using context, such as a user's location, affiliation, and time, to automatically improve the quality of information brokering and delivery. Augmenting an explicit client request with contextual clauses can automatically prioritize, order, and prune information so that the most useful and highest quality among the information available is delivered first. The paper provides techniques for augmenting client requests with context, techniques for combining multiple contextual aspects, and experiments evaluating the efficacy and performance of those techniques. © IFIP International Federation for Information Processing 2009.