Identifying critical features for formative essay feedback with artificial neural networks and backward elimination

Mohsin Abbas*, Peter van Rosmalen, Marco Kalz

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingAcademicpeer-review

Abstract

For predicting and improving the quality of essays, text analytic metrics (surface, syntactic, morphological and semantic features) can be used to provide formative feedback to the students. In this study, the intent was to find a small number of features that exhibit a fair proxy of the scores given by the human raters. Using an existing corpus and a text analysis tool for the Dutch language, a large number of features were extracted. Artificial neural networks, Levenberg Marquardt algorithm and backward elimination were used to reduce the number of extracted features automatically. Irrelevant features were eliminated based on the inter-rater agreement between predicted and human scores calculated using Cohen’s Kappa. By using our algorithm, the number of features in this study was reduced from 457 to 23. The selected features were grouped into six different categories. Of these categories, we believe that the features present in the groups “Word Difficulty” and “Lexical Diversity” are most useful for providing automated formative feedback to the students. The approach presented in this research paper is the first step towards our ultimate goal of providing meaningful formative feedback to the students for enhancing their writing skills and capabilities.
Original languageEnglish
Title of host publicationTransforming Learning with Meaningful Technologies
Subtitle of host publication14th European Conference on Technology Enhanced Learning, EC-TEL 2019, Delft, The Netherlands, September 16–19, 2019, Proceedings
EditorsMaren Scheffel, Julien Broisin, Viktoria Pammer-Schindler, Andrea Ioannou, Jan Schneider
Place of PublicationCham
PublisherSpringer
Chapter30
Pages396-408
Number of pages13
ISBN (Electronic)9783030297367
ISBN (Print)9783030297350
DOIs
Publication statusPublished - 9 Sep 2019
Event14th European Conference on Technology Enhanced Learning: Transforming Learning With Meaningful Technologies - The Leiden-Delft-Erasmus Center for Education and Learning , Delft, Netherlands
Duration: 16 Sep 201919 Sep 2019
Conference number: 2019
http://www.ec-tel.eu
http://www.ec-tel.eu/index.php?id=918

Publication series

SeriesLecture Notes in Computer Science
Volume11722
ISSN0302-9743

Conference

Conference14th European Conference on Technology Enhanced Learning
Abbreviated titleEC-TEL 2019
CountryNetherlands
CityDelft
Period16/09/1919/09/19
Internet address

Fingerprint

Students
Neural networks
Feedback
Syntactics
Semantics

Keywords

  • Formative feedback
  • Natural Language Processing
  • Neural Networks
  • Backward Elimination
  • Dimensionality reduction
  • Feature selection

Cite this

Abbas, M., van Rosmalen, P., & Kalz, M. (2019). Identifying critical features for formative essay feedback with artificial neural networks and backward elimination. In M. Scheffel, J. Broisin, V. Pammer-Schindler, A. Ioannou, & J. Schneider (Eds.), Transforming Learning with Meaningful Technologies: 14th European Conference on Technology Enhanced Learning, EC-TEL 2019, Delft, The Netherlands, September 16–19, 2019, Proceedings (pp. 396-408). Cham: Springer. Lecture Notes in Computer Science, Vol.. 11722 https://doi.org/10.1007/978-3-030-29736-7_30
Abbas, Mohsin ; van Rosmalen, Peter ; Kalz, Marco. / Identifying critical features for formative essay feedback with artificial neural networks and backward elimination. Transforming Learning with Meaningful Technologies: 14th European Conference on Technology Enhanced Learning, EC-TEL 2019, Delft, The Netherlands, September 16–19, 2019, Proceedings. editor / Maren Scheffel ; Julien Broisin ; Viktoria Pammer-Schindler ; Andrea Ioannou ; Jan Schneider. Cham : Springer, 2019. pp. 396-408 (Lecture Notes in Computer Science, Vol. 11722).
@inproceedings{5c53a8d22f32418284c94c1a575d6b17,
title = "Identifying critical features for formative essay feedback with artificial neural networks and backward elimination",
abstract = "For predicting and improving the quality of essays, text analytic metrics (surface, syntactic, morphological and semantic features) can be used to provide formative feedback to the students. In this study, the intent was to find a small number of features that exhibit a fair proxy of the scores given by the human raters. Using an existing corpus and a text analysis tool for the Dutch language, a large number of features were extracted. Artificial neural networks, Levenberg Marquardt algorithm and backward elimination were used to reduce the number of extracted features automatically. Irrelevant features were eliminated based on the inter-rater agreement between predicted and human scores calculated using Cohen’s Kappa. By using our algorithm, the number of features in this study was reduced from 457 to 23. The selected features were grouped into six different categories. Of these categories, we believe that the features present in the groups “Word Difficulty” and “Lexical Diversity” are most useful for providing automated formative feedback to the students. The approach presented in this research paper is the first step towards our ultimate goal of providing meaningful formative feedback to the students for enhancing their writing skills and capabilities.",
keywords = "Formative feedback, Natural Language Processing, Neural Networks, Backward Elimination, Dimensionality reduction, Feature selection",
author = "Mohsin Abbas and {van Rosmalen}, Peter and Marco Kalz",
year = "2019",
month = "9",
day = "9",
doi = "10.1007/978-3-030-29736-7_30",
language = "English",
isbn = "9783030297350",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "396--408",
editor = "Maren Scheffel and Julien Broisin and Viktoria Pammer-Schindler and Andrea Ioannou and Jan Schneider",
booktitle = "Transforming Learning with Meaningful Technologies",

}

Abbas, M, van Rosmalen, P & Kalz, M 2019, Identifying critical features for formative essay feedback with artificial neural networks and backward elimination. in M Scheffel, J Broisin, V Pammer-Schindler, A Ioannou & J Schneider (eds), Transforming Learning with Meaningful Technologies: 14th European Conference on Technology Enhanced Learning, EC-TEL 2019, Delft, The Netherlands, September 16–19, 2019, Proceedings. Springer, Cham, Lecture Notes in Computer Science, vol. 11722, pp. 396-408, 14th European Conference on Technology Enhanced Learning, Delft, Netherlands, 16/09/19. https://doi.org/10.1007/978-3-030-29736-7_30

Identifying critical features for formative essay feedback with artificial neural networks and backward elimination. / Abbas, Mohsin; van Rosmalen, Peter; Kalz, Marco.

Transforming Learning with Meaningful Technologies: 14th European Conference on Technology Enhanced Learning, EC-TEL 2019, Delft, The Netherlands, September 16–19, 2019, Proceedings. ed. / Maren Scheffel; Julien Broisin; Viktoria Pammer-Schindler; Andrea Ioannou; Jan Schneider. Cham : Springer, 2019. p. 396-408 (Lecture Notes in Computer Science, Vol. 11722).

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingAcademicpeer-review

TY - GEN

T1 - Identifying critical features for formative essay feedback with artificial neural networks and backward elimination

AU - Abbas, Mohsin

AU - van Rosmalen, Peter

AU - Kalz, Marco

PY - 2019/9/9

Y1 - 2019/9/9

N2 - For predicting and improving the quality of essays, text analytic metrics (surface, syntactic, morphological and semantic features) can be used to provide formative feedback to the students. In this study, the intent was to find a small number of features that exhibit a fair proxy of the scores given by the human raters. Using an existing corpus and a text analysis tool for the Dutch language, a large number of features were extracted. Artificial neural networks, Levenberg Marquardt algorithm and backward elimination were used to reduce the number of extracted features automatically. Irrelevant features were eliminated based on the inter-rater agreement between predicted and human scores calculated using Cohen’s Kappa. By using our algorithm, the number of features in this study was reduced from 457 to 23. The selected features were grouped into six different categories. Of these categories, we believe that the features present in the groups “Word Difficulty” and “Lexical Diversity” are most useful for providing automated formative feedback to the students. The approach presented in this research paper is the first step towards our ultimate goal of providing meaningful formative feedback to the students for enhancing their writing skills and capabilities.

AB - For predicting and improving the quality of essays, text analytic metrics (surface, syntactic, morphological and semantic features) can be used to provide formative feedback to the students. In this study, the intent was to find a small number of features that exhibit a fair proxy of the scores given by the human raters. Using an existing corpus and a text analysis tool for the Dutch language, a large number of features were extracted. Artificial neural networks, Levenberg Marquardt algorithm and backward elimination were used to reduce the number of extracted features automatically. Irrelevant features were eliminated based on the inter-rater agreement between predicted and human scores calculated using Cohen’s Kappa. By using our algorithm, the number of features in this study was reduced from 457 to 23. The selected features were grouped into six different categories. Of these categories, we believe that the features present in the groups “Word Difficulty” and “Lexical Diversity” are most useful for providing automated formative feedback to the students. The approach presented in this research paper is the first step towards our ultimate goal of providing meaningful formative feedback to the students for enhancing their writing skills and capabilities.

KW - Formative feedback

KW - Natural Language Processing

KW - Neural Networks

KW - Backward Elimination

KW - Dimensionality reduction

KW - Feature selection

U2 - 10.1007/978-3-030-29736-7_30

DO - 10.1007/978-3-030-29736-7_30

M3 - Conference article in proceeding

SN - 9783030297350

T3 - Lecture Notes in Computer Science

SP - 396

EP - 408

BT - Transforming Learning with Meaningful Technologies

A2 - Scheffel, Maren

A2 - Broisin, Julien

A2 - Pammer-Schindler, Viktoria

A2 - Ioannou, Andrea

A2 - Schneider, Jan

PB - Springer

CY - Cham

ER -

Abbas M, van Rosmalen P, Kalz M. Identifying critical features for formative essay feedback with artificial neural networks and backward elimination. In Scheffel M, Broisin J, Pammer-Schindler V, Ioannou A, Schneider J, editors, Transforming Learning with Meaningful Technologies: 14th European Conference on Technology Enhanced Learning, EC-TEL 2019, Delft, The Netherlands, September 16–19, 2019, Proceedings. Cham: Springer. 2019. p. 396-408. (Lecture Notes in Computer Science, Vol. 11722). https://doi.org/10.1007/978-3-030-29736-7_30