An Algorithm For Gaze Region Estimation On Web Pages

Accurate gaze region estimation on the web is important for the purpose of placing marketing advertisements in web pages and monitoring authenticity of users response in web forms. To identify gaze region on the web we need cheap less technical and non-laboratory setup that minimize user interruptions. This study uses the mouse and keyboard as input devices to collect web usage data. The data collected is categorized into mouse activities keyboard activities and scroll activities. A regression model is then used to estimate gaze region from the categorized data. The results indicate that although mouse cursor data is generated throughout the user session the use of data from keyboard presses and scroll from both keyboard and mouse improves the accuracy of identified gaze regions in web usage data.