Desktop Context Detection Using Implicit Feedback

Abstract

The personal information stored on the desktop usually reaches huge dimensions nowadays. Its handling is even more difficult, taking into account complex environments and tasks we work with. An efficient method of identifying the present working context would mean an easier management of the needed resources. In this paper we propose a new way of identifying desktop usage contexts, based upon a distance between documents, which also takes into account their access timestamps. We investigate and compare our technique with traditional term vector clustering, our initial experiments showing promising results with our proposed approach.