Feature Group Importance for Automated Essay Scoring

Jih Soong Tan*, Ian K. T. Tan

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)


One of the challenges in essay scoring is that it is highly subjective to the human graders. There have been numerous research projects conducted on improving computerised Automated Essay Scoring (AES). AES systems generally rely on hand-crafted linguistic features to construct a classification model for essay scoring. The majority of the AES systems’ classification algorithm inputs are based on three main feature groups; lexical, grammatical, and semantic feature groups. This paper presents an empirical study to explore the influence of each feature group on the performance of AES classification models based on a general approach of the AES system. The results uncovered that the grammatical and semantic feature groups are lacking due to their poor performance and typical over-fitting of the classification models when using the features in the feature group.

Original languageEnglish
Title of host publicationMulti-disciplinary Trends in Artificial Intelligence. MIWAI 2021
EditorsPhatthanaphong Chomphuwiset, Junmo Kim, Pornntiwa Pawara
Number of pages13
ISBN (Electronic)9783030802530
ISBN (Print)9783030802523
Publication statusPublished - 2021
Event14th International Conference on Multi-disciplinary Trends in Artificial Intelligence 2021 - Virtual, Online
Duration: 2 Jul 20213 Jul 2021

Publication series

NameLecture Notes in Computer Science
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference14th International Conference on Multi-disciplinary Trends in Artificial Intelligence 2021
Abbreviated titleMIWAI 2021
CityVirtual, Online


  • ASAP
  • Auto essay scoring
  • EASE
  • Features
  • Importance

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science


Dive into the research topics of 'Feature Group Importance for Automated Essay Scoring'. Together they form a unique fingerprint.

Cite this