Predicting misreadings from gaze in children with reading difficulties

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

We present the first work on predicting reading mistakes in children with reading difficulties based on eye-tracking data from real-world reading teaching. Our approach employs sev-eral linguistic and gaze-based features to in-form an ensemble of different classifiers, in-cluding multi-task learning models that let us transfer knowledge about individual readers to attain better predictions. Notably, the data we use in this work stems from noisy readings in the wild, outside of controlled lab condi-tions. Our experiments show that despite the noise and despite the small fraction of mis-readings, gaze data improves the performance more than any other feature group and our models achieve good performance. We further show that gaze patterns for misread words do not fully generalize across readers, but that we can transfer some knowledge between readers using multitask learning at least in some cases. Applications of our models include partial au-tomation of reading assessment as well as per-sonalized text simplification.
Original languageEnglish
Title of host publicationProceedings of the Thirteenth Workshop on Innovative Use of NLP for Building Educational Applications,
PublisherAssociation for Computational Linguistics
Publication date2018
Pages29-34
DOIs
Publication statusPublished - 2018
Event13h Workshop on Innovative Use of NLP for Building Educational Applications - New Orleans, United States
Duration: 5 Jun 20185 Jun 2018

Workshop

Workshop13h Workshop on Innovative Use of NLP for Building Educational Applications
LandUnited States
ByNew Orleans
Periode05/06/201805/06/2018

ID: 217116273