Readability Analysis Based on Cognitive Assessment Using Physiological Sensing

Abstract

For a given individual, the readability of a content determines the amount of cognitive load experienced during the reading. Among many physiological parameters, the brain signals provide direct information about the cognitive load. In this paper, for a silent reading scenario, electroencephalogram (EEG) signal and pupillary response are used to address both global and localized analysis of a content. The global analysis is done using the EEG data collected from a target group of readers on whether the overall content is easy or difficult to understand. The entropy analysis of left temporal brain signals is found to grossly differentiate the texts. Detailed analysis is done using the pupil dilation to localize certain difficult segments in the content. The oscillatory behavior of the pupil dilation reveals detailed characteristics of the effect of the texts on each individual. To carry out the experiment, two types (easy and difficult) of textual contents are considered which are calibrated using standard natural language processing parameters. The proposed work presents a methodology for global and detailed analysis of a content which helps in preparing the reading materials. Furthermore, the methodology provides the insights into the reading characteristics of the individuals.