Online Assessment of Students' Text Comprehension: Explorations into the Automated Scoring of Constructed Responses

EasyChair Preprint 3870

Abstract

Effective computerized reading comprehension strategy training requires an ability to provided automated an accurate to students as they answer open-ended comprehension question. This study explored different approaches to machine scoring as part of a larger research and development project. Two units involving 4 open-ended questions were used for initial testing. A comparison of simple frequency and deep learning techniques suggest that the latter have more potential to provide accurate feedback on response correctness.

Keyphrases: Question Answering, assessment, deep learning, machine learning, reading comprehension

@booklet{EasyChair:3870,
  author    = {Jean-François Rouet and Peter Hastings and Mônica Macedo-Rouet and Anna Potocki and M. Anne Britt},
  title     = {Online Assessment of Students' Text Comprehension: Explorations into the Automated Scoring of Constructed Responses},
  howpublished = {EasyChair Preprint 3870},
  year      = {EasyChair, 2020}}