Title: Assessment of reading material using sensor data

Authors: Aniruddha Sinha; Sanjoy Kumar Saha; Anupam Basu

Addresses: TCS Research and Innovation Lab, Tata Consultancy Services, Kolkata, India ' Department of Computer Science and Engineering, Jadavpur University, Kolkata, India ' Department of Computer Science and Engineering, Indian Institute of Technology, Kharagpur, India; National Institute of Technology, Dugrapur, India

Abstract: Reading is characterised by a sequence of complex processing in the brain. The experienced cognitive load and engagement level play a major role in the assimilation of content. In this paper, for the evaluation of a textual content, we use electroencephalogram (EEG) for brain signals and eyetracker for the eyegaze data. Experiment is done with two types (easy and difficult) of textual contents which are benchmarked using standard parameters of natural language processing. Features on cognitive load and engagement index extracted from alpha and beta frequency bands of EEG are found to be discriminative from left-temporal and right-prefrontal lobes respectively. Statistical features, derived from the shift in eyegaze fixations from the current line to the adjacent lines, are found to be discriminative. A difficulty score is computed using a novel mapping function derived from the mixture of two partial sigmoid. This enables more objective comparison of two contents and helps in finding differences in individual reading skills.

Keywords: textual content; eye-gaze; electroencephalogram; EEG; cognitive load; sigmoid.

DOI: 10.1504/IJAAC.2022.125283

International Journal of Automation and Control, 2022 Vol.16 No.5, pp.519 - 546

Received: 25 Feb 2019
Accepted: 19 Oct 2019
Published online: 06 Sep 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article