Interruption-sensitive empty result feedback: Rethinking the visual query feedback paradigm for semistructured data

The usability of visual querying schemes for tree and graph-structured data can be greatly enhanced by providing feedback during query construction, but feedback at inopportune times can hamper query construction. In this paper, we rethink the traditional way of providing feedback. We describe a novel vision of interruption-sensitive query feedback where relevant notifications are delivered quickly but at an appropriate moment when the mental workload of the user is low. Though we focus on one class of query feedback, namely empty result detection, where a user is notified when a partially constructed visual query yields an empty result, our new paradigm is applicable to other kinds of feedback. We present a framework called iSERF that bridges the classical database problem of empty-result detection with intelligent notification management from the domains of HCI and psychology. Instead of immediate notification, iSERF considers the structure of query formulation tasks and breakpoints when reasoning about when to notify the user. We present an HCI-inspired model to quantify the performance bounds that iSERF must abide by for checking for an empty result in order to ensure interruption-sensitive notification at optimal breakpoints. We implement this framework in the context of visual XML query formulation and highlight its effectiveness empirically.