Date of Award
Master of Science (MS)
Most smartphone applications are unaware that users feel frustrated by a bug, an error, or a usability problem. Could a smartphone be “smart” enough to detect that its user became frustrated by something? A pilot study was conducted as proof of concept, showing that sensor readings collected in the background from a smartphone could be used to detect user frustration after the onset of a frustrating event due to a bug or a usability problem. Twenty-one participants were asked to perform a series of multitasking tasks, during which errors were purposely introduced to frustrate them. Sensor data, including motion, touch, and camera, were collected and used to train a binary classifier that is able to detect frustration with reasonable accuracy when merging data from different modalities (motion sensors and touch gestures).
Vasiete Allas, Esther, "Detecting User Frustration From Smartphone Sensors: A Multimodal Classification Approach" (2015). Electrical Engineering Graduate Theses & Dissertations. 10.