In pervasive environments, several devices are present at any given time; many of these perform the same, or similar, tasks. Thus, when a user wants to perform a task, he can choose the device best suited for it. Though this may seem to be an advantage at first glance, in practice, it is an inconvenience because it forces the user to make an often-difficult choice between apparently similar devices.

In this paper, the authors overcome this burden by presenting a dynamic user task composition method based on user preferences: in practice, the user just declares a set of preferences, and then devices interact automatically with each other, following these preferences. The paper is quite thorough. After an introduction explaining what the paper is about, and a quick presentation of previous and related work, it presents modeling methods for both user preferences and tasks. It then describes a practical Java implementation of this method, which in the future might be incorporated into an open-source cloud environment.

This work is a research paper: a conceptual framework outlining advantages and disadvantages of a given approach. At this stage, it proposes algorithms, not working solutions or commercial products; even the Java prototype serves only validation purposes. Consequently, the main audience for this paper is the research community in the field of autonomous and adaptive systems, which is presented with new concepts and advancements.