Detection of Low Satisfaction Web Pages with Discriminant Analysis using Data of User's Behavior

Abstract

The purpose of this work is to reduce the target pages of the web usability evaluation by usability testing. The target pages will reduce by detecting low satisfaction web pages. We analyze users ' interaction data for detecting low satisfaction web pages using variable selection. Analysis results, detectable interaction data are browsing time, moving distance of gazing points and wheel rolling. We apply statistical discriminant analysis to these interaction data and detect low usability web pages by classification between low usability group and others. Analysis result, we get mathematical line between evaluation target pages and low usability rate. Evaluators can determine evaluation target pages by using mathematical line of low usability group.