Transparency strategies to help users handle system errors

This study analyzes the impact of transparency strategies on the quality of user responses following system rejections and misunderstandings. The aim of transparency is to make the system “visible” according to user needs, i.e. when the user's system knowledge is not sufficiently comprehensive and/or correct to successfully achieve the current goal. Since system knowledge depends on users' expertise, system expectations and system error rank, transparency is treated in this study as an adaptable and adaptive feature. Adaptable and adaptive transparency strategies were applied to TRAVELS, a spoken dialogue system which enables users to obtain plane and train schedules over the phone. Based on a partial Wizard-of-Oz simulation, an empirical assessment indicates the extent to which these transparency strategies can help users to respond appropriately to system errors.