Facilitating user involvement in intelligent interactive systems

点击翻译按钮获取中文摘要

Intelligent interactive systems embed techniques from the field of artificial intelligence (AI) to tailor their human-computer interactions to individual user needs. Given the right underlying reasoning mechanisms, intelligent interactive systems have great potential to improve users' experience as they interact with complex applications and information spaces both qualitatively (e.g., user satisfaction) and quantitatively (e.g., task efficiency). Their success, however, relies not only on their ability to model and reason about relevant user and contextual factors, but also on the manner in which the intelligence is delivered, or in other words, the interaction design. Of particular importance in the interaction design is providing sufficient opportunities for user involvement. The proposed research will examine issues such as how to make the system's intelligent behaviour understandable, how to divide the labour between the user and system so that their respective strengths are leveraged, and under what circumstances an intelligent interactive system should provide proactive advice. These issues will be examined through two complementary approaches: i) using qualitative methods, such as semi-structured interviews, to understand how users perceive and use deployed intelligent interaction systems; and ii) creating and systematically evaluating research prototypes to assess the impact of aspects of their design. A better understanding of how to design intelligent interactive systems will benefit both industrial and research communities. There are millions users in Canada and worldwide interacting with technology that could be made substantially more efficient and effective from an individual user's perspective given the right intelligent support, including search engines, e-commerce websites and productivity applications (e.g., word processors). Within the research community, this work will help those designing underlying AI reasoning mechanisms to ensure that poor usability is not a confounding factor in any validation work involving human participants. A second contribution of this work will be a strong methodological basis with which to build and study intelligent interactive systems.