Identifying Reading Styles from E-book Log Data

Abstract

In this paper, a model for identifying e-book reading style is proposed and applied onto a learning log dataset. Learning log data available as non-structured data source is processed to identify patterns of reading exhibited by users using three main structures: reading sessions, reads and passages. These structures are used to extract information on users’ reading style to be used as part of user modeling process. The proposed model is applied on a set of log data generated by university students during one semester of digital resource use. The findings show students adopt predominantly receptive reading style, while responsive style occurs rarely. Further analysis revealed no significant relationships between reading style variables and student academic success for the Architecture course indicating the variables of responsive and receptive reading bring new information as part of user modeling.