Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics

Min Chi, Pamela Jordan, Kurt VanLehn, Moses Hall

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

15 Citations (Scopus)

Abstract

Given the subtlety of tutorial tactics, identifying effective pedagogical tactical rules from human tutoring dialogues and implementing them for dialogue tutoring systems is not trivial. In this work, we used reinforcement learning (RL) to automatically derive pedagogical tutoring dialog tactics. Past research has shown that the choice of the features significantly affects the effectiveness of the learned tactics. We defined a total of 18 features which we classified into four types. First, we compared five feature selection methods and overall upper-bound method seems to be most efficient. Then we compared the four types of features and found that temporal situation and autonomy related features are significantly more relevant and effective to tutorial decisions than either performance or situation related features.

Original languageEnglish (US)
Title of host publicationEducational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings
Pages258-265
Number of pages8
StatePublished - 2008
Externally publishedYes
Event1st International Conference on Educational Data Mining, EDM 2008 - Montreal, QC, Canada
Duration: Jun 20 2008Jun 21 2008

Other

Other1st International Conference on Educational Data Mining, EDM 2008
CountryCanada
CityMontreal, QC
Period6/20/086/21/08

Fingerprint

Reinforcement learning
Feature extraction

ASJC Scopus subject areas

  • Computer Science Applications
  • Information Systems

Cite this

Chi, M., Jordan, P., VanLehn, K., & Hall, M. (2008). Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics. In Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings (pp. 258-265)

Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics. / Chi, Min; Jordan, Pamela; VanLehn, Kurt; Hall, Moses.

Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings. 2008. p. 258-265.

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

Chi, M, Jordan, P, VanLehn, K & Hall, M 2008, Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics. in Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings. pp. 258-265, 1st International Conference on Educational Data Mining, EDM 2008, Montreal, QC, Canada, 6/20/08.
Chi M, Jordan P, VanLehn K, Hall M. Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics. In Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings. 2008. p. 258-265
Chi, Min ; Jordan, Pamela ; VanLehn, Kurt ; Hall, Moses. / Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics. Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings. 2008. pp. 258-265
@inproceedings{641e12fdbc394faea0349972ddbafbde,
title = "Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics",
abstract = "Given the subtlety of tutorial tactics, identifying effective pedagogical tactical rules from human tutoring dialogues and implementing them for dialogue tutoring systems is not trivial. In this work, we used reinforcement learning (RL) to automatically derive pedagogical tutoring dialog tactics. Past research has shown that the choice of the features significantly affects the effectiveness of the learned tactics. We defined a total of 18 features which we classified into four types. First, we compared five feature selection methods and overall upper-bound method seems to be most efficient. Then we compared the four types of features and found that temporal situation and autonomy related features are significantly more relevant and effective to tutorial decisions than either performance or situation related features.",
author = "Min Chi and Pamela Jordan and Kurt VanLehn and Moses Hall",
year = "2008",
language = "English (US)",
isbn = "9780615306292",
pages = "258--265",
booktitle = "Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings",

}

TY - GEN

T1 - Reinforcement learning-based feature selection for developing pedagogically effective tutorial dialogue tactics

AU - Chi, Min

AU - Jordan, Pamela

AU - VanLehn, Kurt

AU - Hall, Moses

PY - 2008

Y1 - 2008

N2 - Given the subtlety of tutorial tactics, identifying effective pedagogical tactical rules from human tutoring dialogues and implementing them for dialogue tutoring systems is not trivial. In this work, we used reinforcement learning (RL) to automatically derive pedagogical tutoring dialog tactics. Past research has shown that the choice of the features significantly affects the effectiveness of the learned tactics. We defined a total of 18 features which we classified into four types. First, we compared five feature selection methods and overall upper-bound method seems to be most efficient. Then we compared the four types of features and found that temporal situation and autonomy related features are significantly more relevant and effective to tutorial decisions than either performance or situation related features.

AB - Given the subtlety of tutorial tactics, identifying effective pedagogical tactical rules from human tutoring dialogues and implementing them for dialogue tutoring systems is not trivial. In this work, we used reinforcement learning (RL) to automatically derive pedagogical tutoring dialog tactics. Past research has shown that the choice of the features significantly affects the effectiveness of the learned tactics. We defined a total of 18 features which we classified into four types. First, we compared five feature selection methods and overall upper-bound method seems to be most efficient. Then we compared the four types of features and found that temporal situation and autonomy related features are significantly more relevant and effective to tutorial decisions than either performance or situation related features.

UR - http://www.scopus.com/inward/record.url?scp=84857469933&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84857469933&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9780615306292

SP - 258

EP - 265

BT - Educational Data Mining 2008 - 1st International Conference on Educational Data Mining, Proceedings

ER -