Neural machine translation for automated feedback on children's early-stage writing

Jonas Vestergaard Jensen, Mikkel Jordahn, Michael Riis Andersen

Keywords: natural language processing, neural machine translation, sequence-to-sequence, robust likelihood, automated feedback, computational linguistics

Abstract: In this work, we address the problem of assessing and constructing feedback for early-stage writing automatically using machine learning. Early-stage writing is typically vastly different from conventional writing due to phonetic spelling and lack of proper grammar, punctuation, spacing etc. Consequently, early-stage writing is highly non-trivial to analyze using common linguistic metrics. We propose to use sequence-to-sequence models for translating early-stage writing by students into conventional writing, which allows the translated text to be analyzed using linguistic metrics. Furthermore, we propose a novel robust likelihood to mitigate the effect of label noise in the dataset. We investigate the proposed methods using a set of numerical experiments and demonstrate that the conventional text can be predicted with high accuracy.

Submission Number: 22