Arroyo, E. "Mediating Disruption in Human-Computer Interaction from Implicit Metrics of Attention"

Abstract

Multitasking environments cause people to be interrupted constantly, often disrupting their ongoing activities and impeding reaching their goals. This thesis presents a disruption reducing approach designed to support the user's goals and optimize productivity that is based on a model of the user's receptivity to an interruption. The model uses knowledge of the interruption content, context and priority of the task(s) in progress, user actions and goal-related concepts to mediate interruptions.

The disruption management model is distinct from previous work by the addition of implicit sensors that deduce the interruption content and user context to help determine when an interruption will disrupt an ongoing activity. Domain-independent implicit sensors include mouse and keyboard behaviors, and goal-related concepts extracted from the user documents. The model also identifies the contextual relationship between interruptions and user goals as an important factor in how interruptions are controlled. The degree to which interruptions are related to the user goal determines how those interruptions will be received.

We tested and evolved the model in various cases and showed significant improvement in both productivity and satisfaction. A disruption manager application controls interruptions on common desktop computing activities, such as web browsing and instant messaging. The disruption manager demonstrates that mediating interruptions by supporting the user goals can improve performance and overall productivity. Our evaluation shows an improvement in success of over 25% across prioritization conditions for real life computing environments.

Goal priority and interruption relevance play an important role in the interruption decision process and several experiments these factors on people's reactions and availability to interruptions, and overall performance. These experiments demonstrate that people recognize the potential benefits of being interrupted and adjust their susceptibility to interruptions during highly prioritized tasks.

The outcome of this research includes a usable model that can be extended to tasks as diverse as driving an automobile and performing computer tasks. This thesis supports mediating technologies that will recognize the value of communication and control interruptions so that people are able to maintain concentration amidst their increasingly busy lifestyles