Loading...
Please wait, while we are loading the content...
Similar Documents
Improving Understanding , Trust , and Control with Intelligibility in Context-Aware Applications
| Content Provider | Semantic Scholar |
|---|---|
| Author | Lim, Brian |
| Copyright Year | 2011 |
| Abstract | Context-aware applications can facilitate people as they carry out their daily tasks. These applications can use a suite of sensors to detect what is happening in the environment and with the user. They can then infer the user intention. This way, they try to understand the contexts of the situation, and consequently act to provide services. For example, a smart phone can recognize that you are in a conversation, and suppress any incoming messages during this period. To minimize obtrusiveness and allow users to focus primarily on their tasks, context-aware applications perform sensing implicitly without explicitly informing users. Furthermore, to better understand the contexts of users in their physical and social environments, contextaware applications are using increasingly complex mechanisms to infer these contexts (e.g., by using machine learning algorithms). This implicit sensing and complex inference can remain invisible when the applications work well and as expected, but become a mystery when the applications behave inappropriately or unexpectedly. In such cases, the lack of understanding of these applications can lead users to mistrust, misuse it, or abandon them altogether. To counter this, contextaware applications should be intelligible, capable of generating explanations of their behavior. This thesis proposes to investigate how to provide intelligibility in context-aware applications, and evaluate its usefulness to improve user understanding, trust, and control. We explored what explanation types users are interested in for different context-aware applications under various circumstances. We provided explanations in terms of questions that users would ask, such as why did it do X, why did it not do Y, what if I did W, what will it do, how can I get the application to do Y? Early evaluation found why and why not explanation types most effective, among these four explanation types, to improve understanding and trust in intelligent context-aware applications. We have developed a toolkit to help developers implement intelligibility in their context-aware applications, such that they can automatically generate explanations. However, presenting explanations remained an open design exercise, so we explored usability issues of and design principles for providing intelligibility in a prototype of a mobile context-aware application. Having developed technical and design support for intelligibility in context-aware applications, we next seek to further evaluate intelligibility on improving understanding, trust, and control. First, we shall explore the limits of the helpfulness of intelligibility given the uncertainty of the application. Due to the scrutable and transparent nature of intelligibility, we hypothesize that, while intelligibility helps users better appreciate a highly certain application, it would also expose the inadequacy of a highly uncertain application. We propose to investigate the impact on user impression on context-aware applications with high and low certainty. Next, we shall explore the symbiotic relationship between intelligibility and control. With the greater understanding from intelligibility, users should be able to better control and configure context-aware applications. Conversely, being empowered with the capability to control the application, users would be able to better understand and appreciate intelligibility. So we propose to investigate how the combination of intelligibility and controllability improves understanding, trust, and control of context-aware applications. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.brianlim.net/wordpress/wp-content/uploads/2011/04/brianlim-thesis-proposal.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
