A person seeking someone else's attention is normally able to quickly assess how interruptible they are. This assessment allows for behavior we perceive as natural, socially appropriate, or simply polite. On the other hand, today's computer systems are almost entirely oblivious to the human world they operate in, and typically have no way to take into account the interruptibility of the user. This paper presents a Wizard of Oz study exploring whether, and how, robust sensor-based predictions of interruptibility might be constructed, which sensors might be most useful to such predictions, and how simple such sensors might be. The study simulates a range of possible sensors through human coding of audio and video recordings. Experience sampling is used to simultaneously collect randomly distributed self-reports of interruptibility. Based on these simulated sensors, we construct statistical models predicting human interruptibility and compare their predictions with the collected self-report data. The results of these models, although covering a demographically limited sample, are very promising, with the overall accuracy of several models reaching about 78%. Additionally, a model tuned to avoiding unwanted interruptions does so for 90% of its predictions, while retaining 75% overall accuracy.

User attention is a scarce resource, and users are susceptible to interruption overload. Systems do not reason about the effects of interrupting a user during a task sequence. In this study, we measure effects of interrupting a user at different moments within task execution in terms of task performance, emotional state, and social attribution. Task models were developed using event perception techniques, and the resulting models were used to identify interruption timings based on a user’s predicted cognitive load. Our results show that different interruption moments have different impacts on user emotional state and positive social attribution, and suggest that a system could enable a user to maintain a high level of awareness while mitigating the disruptive effects of interruption. We discuss implications of these results for the design of an attention manager. Categories & Subject Descriptors H.5.2 [Information Interfaces and Presentation]: User Interfaces — evaluation/methodology, user-centered

At first glance it seems absurd that busy people doing important jobs should want their computers to interrupt them. Interruptions are disruptive and people need to concentrate to make good decisions. However, successful job performance also frequently depends on people's abilities to (a) constantly monitor their dynamically changing information environments, (b) collaborate and communicate with other people in the system, and (c) supervise background autonomous services. These critical abilities can require people to simultaneously query a large set of information sources, continuously monitor for important events, and respond to and communicate with other human operators. Automated monitoring

Designing and evaluating notification systems represents an emerging challenge in the study of human--computer interaction. Users rely on notification systems to present potentially interruptive information in an efficient and effective manner to enable appropriate reaction and comprehension. Little is known about the effects of these systems on ongoing computer tasks. As the research community strives to understand information design suitable for opposing usage goals, few existing efforts lend themselves to extensibility.

The computer and communication systems that office workers currently use tend to interrupt at inappropriate times or unduly demand attention because they have no way to determine when an interruption is appropriate. Sensor-based statistical models of human interruptibility offer a potential solution to this problem. Prior work to examine such models has primarily reported results related to social engagement, but it seems that task engagement is also important. Using an approach developed in our prior work on sensor-based statistical models of human interruptibility, we examine task engagement by studying programmers working on a realistic programming task. After examining many potential sensors, we implement a system to log low-level input events in a development environment. We then automatically extract features from these low-level event logs and build a statistical model of interruptibility. By correctly identifying situations in which programmers are non-interruptible and minimizing cases where the model incorrectly estimates that a programmer is non-interruptible, we can support a reduction in costly interruptions while still allowing systems to convey notifications in a timely manner. Author Keywords Situationally appropriate interaction, managing human attention, sensor-based interfaces, context-aware computing, machine learning, interruptibility.

This research adopted a model of goal activation to study the mechanisms underlying interrupted task performance. The effects of interruption timing, type of interruption, and age on task time and primary task resumption time were explored under conditions in which attention was switched back and forth between two tasks, much as when drivers shift attention between attending to the road and to an in-vehicle task. The timing of interruptions had a significant impact on task resumption times, indicating that the most costly time to interrupt task performance is during the middle of a task. However, this effect was overshadowed by age-related performance decrements for older participants. Interruptions that prevented strategic rehearsal of goals resulted in longer resumption times as compared with interruptions that allowed rehearsal. Actual or potential applications of this research include the design of in-vehicle device user interfaces, the timing of invehicle messages, and current metrics for assessing driver distraction.

This article provides a first look at an extensible philosophy for studying other instances of multitasking or collaborative performance. We argue that the models and framework presented here will improve the HCI community's ability to classify and evaluate existing and emerging notification systems, as well as to catalog information and interaction design guidelines and lessons learned in a cohesive, collective manner. In the next section, we present a more thorough overview of notification systems appearing in recent literature and itemize general user goals, providing motivation and background material for the model we present in Section 3

We elaborate a proposal for capturing, extending, and reusing design knowledge gleaned through usability testing. The proposal is specifically targeted to address interface design for notification systems, but its themes can be generalized to any constrained and well-defined genre of interactive system design. We reiterate arguments for and against using critical parameters to characterize user goals and usability artifacts. Responding to residual arguments, we suggest that clear advantages for research cohesion, design knowledge reuse, and HCI education are possible if several challenges are overcome. As a first step, we recommend a slight variation to the concept of a critical parameter, which would allow both abstract and concrete knowledge representation. With this concept, we demonstrate a feasible approach by introducing equations that elaborate and allow evolution of notification system critical parameters, which is made operational with a variety of usability evaluation instruments. A case study illustrates how one general instrument allowed system designs to be meaningfully compared and resulted in valuable inferences for interface reengineering. Broad implications and conclusions about this approach will be of interest to others concerned with using critical parameters in interface design, development of notification systems interfaces, or approaches to design rationale and knowledge reuse.

Applications that use sensor-based estimates face a fundamental tradeoff between true positives and false positives when examining the reliability of these estimates, one that is inadequately described by the straightforward notion of accuracy. To address this tradeoff, this paper examines the use of Receiver Operating Characteristic (ROC) curve analysis, a method that has a long history but is under-appreciated in the human computer interaction research community. We present the fundamentals of ROC analysis, the use of the A ' statistic to compute the area under an ROC curve, and the equivalence of A ' to the Wilcoxon statistic. We then present several case studies, framed in the context of our work on human interruptibility, demonstrating how ROC analysis can yield better results than analyses based on accuracy. These case studies compare sensor-based estimates with human performance, optimize a feature selection process for the area under the ROC curve, and examine end-user selection of a desirable tradeoff.

Human perceptual and cognitive abilities are limited resources. Attention is the mechanism used to allocate such resources in the most effective way. Current technologies, in addition to allowing fast access to information and people, should be designed to support human attentional processes on which they impose further strain. This paper analyses the issues related to the design of systems capable of such support: Attention Aware Systems. We introduce the research aimed at understanding and modelling human attentional processes, including perceptual and cognitive processes as studied in cognitive psychology, as well as rhetorical, aesthetic, and social aspects related to attentional mechanisms. We analyse current approaches to the design of Attention Aware Systems along three major features: detection of user's current attentional state, detection and evaluation of possible alternative attentional states, strategies for focus switch or maintenance. Finally, we discuss the most promising research direction for the development of systems capable of supporting human attentional mechanisms.