SHF: Small: Explanation Logic

Computing technology has a rapidly growing impact on people's lives. Therefore, it becomes ever more important for computing systems to explain their behavior and to reassure users why they should trust them. Since the complexity of automated tasks is also steadily increasing, explanations of computation results and decisions will be non-trivial and potentially contain a large number of details, which poses the challenge of delivering the right information at the right level of detail.

To be effective, explanations must dynamically adapt to specific users and situations. This research explores the nature of dynamic, adaptable explanations and how they can effectively help users to gain an understanding of and trust in the behavior of complex computing systems. The project’s novelty is the focus on the dynamic nature of explanations to facilitate flexible, adaptive responses to users, and the project's impacts are more transparent future computing systems, more widely accepted by more people, brought about by methods and tools that affect the design of future software.

The overall goal of this research project is to develop a logical basis for reasoning about understanding in support of explanation systems. Based on the definition of a new modality of understanding (which is different from knowledge and belief), the investigator is developing a dynamic two-agent explanation logic (with user and system as agents), which allows the description of the evolution of understanding.

The research project is producing a dynamic model for explanation logic that is to be the basis for a generic explanation inference algorithm that mirrors the dynamically shrinking set of possible worlds with a correspondingly shrinking set of tailored explanations. In addition, the research is developing concepts for tailorable explanation structures to support the effective communication of explanations that are customized to specific user needs.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.