Assessing question quality using NLP

Kristopher J. Kopp, Amy Johnson, Scott A. Crossley, Danielle McNamara

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

2 Scopus citations

Abstract

An NLP algorithm was developed to assess question quality to inform feedback on questions generated by students within iSTART (an intelligent tutoring system that teaches reading strategies). A corpus of 4575 questions was coded using a four-level taxonomy. NLP indices were calculated for each question and machine learning was used to predict question quality. NLP indices related to lexical sophistication modestly predicted question type. Accuracies improved when predicting two levels (shallow versus deep).

Original languageEnglish (US)
Title of host publicationArtificial Intelligence in Education - 18th International Conference, AIED 2017, Proceedings
EditorsElisabeth Andre, Xiangen Hu, Ma. Mercedes T. Rodrigo, Benedict du Boulay, Ryan Baker
PublisherSpringer Verlag
Pages523-527
Number of pages5
ISBN (Print)9783319614243
DOIs
StatePublished - 2017
Event18th International Conference on Artificial Intelligence in Education, AIED 2017 - Wuhan, China
Duration: Jun 28 2017Jul 1 2017

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10331 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other18th International Conference on Artificial Intelligence in Education, AIED 2017
Country/TerritoryChina
CityWuhan
Period6/28/177/1/17

Keywords

  • Artificial intelligence
  • Educational technology design
  • Intelligent tutoring systems
  • Natural language processing
  • Question classification

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Assessing question quality using NLP'. Together they form a unique fingerprint.

Cite this