Identification of User Confusion in a Web Application

Web applications are daily visited by everyone of us. There is variety of users from novice ones to experts. While filling our needs we might be confused, many times. For example, it might be in the situation when we visit the application for the first time, but confusion might also occurred dealing with application we are experienced with, while trying some new feature or after update. These situations might be problematic. In this work we deal with automatic identification of user confusion in web application.

Proposed method works with interaction data from keyboard and mouse. We focus on identification of right moment to display a guide for user. This guide may consists of hints which will explain how to work with certain web application.

We conduct a user study with 60 participants working on 6 tasks at FIRO tour travel agency’s website. Based on the gathered data we trained a machine learning classifier using logistic regression for real time confusion prediction. Reached results show that we are able to predict the confusion in web application using only mouse interaction data.