國立虎尾科技大學 |

Stealth Assessment of Self-Regulated Learning in Game-Based Learning Environments.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Stealth Assessment of Self-Regulated Learning in Game-Based Learning Environments./
Author:	Sabourin, Jennifer Lynne.
Description:	1 online resource (171 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 75-10, Section: B.
Contained By:	Dissertations Abstracts International75-10B.
Subject:	Educational psychology. -
Online resource:	click for full text (PQDT)
ISBN:	9781303799624

Stealth Assessment of Self-Regulated Learning in Game-Based Learning Environments.
Sabourin, Jennifer Lynne.

Stealth Assessment of Self-Regulated Learning in Game-Based Learning Environments. - 1 online resource (171 pages)

Source: Dissertations Abstracts International, Volume: 75-10, Section: B.

Thesis (Ph.D.)--North Carolina State University, 2013.

Includes bibliographical references

Game-based learning environments have been touted as a powerful mechanism for keeping twenty-first century students motivated and engaged in learning tasks. These environments capitalize on important game features such as autonomy, control, and interactivity to maintain student interest. However, they may introduce new challenges by requiring students to engage in more advanced self-regulatory skills such as a goal-setting and monitoring, complex problem solving, and critical thinking. Students without these skills are unlikely to realize the benefits of game-based learning and may struggle in systems where they are provided with too much autonomy. Consequently, real-time identification of self-regulated learning (SRL) is a problem of critical importance for providing adaptive scaffolding targeted to an individual student's needs. The goal of the research presented in this dissertation is to explore the use of a theoretically grounded machine learning framework for the real-time stealth assessment of SRL skills. In this framework, prominent psychological theories are used to guide the selection and application of machine learning and data mining techniques. Empirical models are trained on a corpus of data from students interacting with a game-based learning environment, CRYSTAL ISLAND. These models are evaluated to determine if the predictive models are benefited by the theoretically grounded framework and if they are also sufficiently robust to offer real-time trace-based assessment of SRL. The work consists of four phases of theory and data-driven investigation: (1) identification of SRL, (2) feature selection and creation, (3) model creation and (4) model evaluation. First, SRL skill measures of particular relevance to the CRYSTAL ISLAND environment are identified using a variety of techniques. Each measure is designed to be representative of one of the three components of SRL: cognition, metacognition and motivation. Students are classified based on their demonstration of these skills. Next, feature selection is used to identify simple personal attribute and in-game actions that are tied to SRL classifications. A differential sequence mining approach is then used to identify patterns of behavior that are indicative of each component of SRL. These patterns are used to create complex event features guided by SRL theory. In the model creation phase, these features are integrated in naive, static and dynamic Bayesian networks with structures that are increasingly informed by process models of SRL. The models are then empirically learned from a corpus of student interaction data. Finally, the models are evaluated to determine the value of the theoretically grounded empirical framework. Specifically, the evaluation seeks to quantify usefulness of the features identified in Phase 2 and the structures devised in Phase 3. Furthermore, the models are evaluated to identify how early into the interaction SRL skill can be accurately assessed and how well the models generalize to unseen populations. Overall, the results indicate that the theoretically-grounded machine learning framework is effective for developing real-time trace based assessment models of SRL. The differential sequence mining approach yields patterns of behavior that are both meaningful and improve the predictive accuracy of the models. Dynamic Bayesian models with temporal structures representing the cyclic nature of SRL outperformed those without this representation. Finally, the framework was effective at predicting the cognitive, metacognitive and motivational components of SRL within the first 20% of the interaction. These results highlight the potential of the framework for guiding the design of real-time trace-based SRL models.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9781303799624Subjects--Topical Terms:

555103
Educational psychology.
Subjects--Index Terms:

Educational data miningIndex Terms--Genre/Form:

554714
Electronic books.

Stealth Assessment of Self-Regulated Learning in Game-Based Learning Environments.
LDR:05255ntm a22004217 4500 001 1147915
005 20240916075429.5
006 m o d
007 cr bn ---uuuuu
008 250605s2013 xx obm 000 0 eng d
020 $a 9781303799624
035 $a (MiAaPQ)AAI3586200
035 $a AAI3586200
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Sabourin, Jennifer Lynne. $3 1473758
245 1 0 $a Stealth Assessment of Self-Regulated Learning in Game-Based Learning Environments.
264 0 $c 2013
300 $a 1 online resource (171 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 75-10, Section: B.
500 $a Publisher info.: Dissertation/Thesis.
500 $a Advisor: Lester, James C.
502 $a Thesis (Ph.D.)--North Carolina State University, 2013.
504 $a Includes bibliographical references
520 $a Game-based learning environments have been touted as a powerful mechanism for keeping twenty-first century students motivated and engaged in learning tasks. These environments capitalize on important game features such as autonomy, control, and interactivity to maintain student interest. However, they may introduce new challenges by requiring students to engage in more advanced self-regulatory skills such as a goal-setting and monitoring, complex problem solving, and critical thinking. Students without these skills are unlikely to realize the benefits of game-based learning and may struggle in systems where they are provided with too much autonomy. Consequently, real-time identification of self-regulated learning (SRL) is a problem of critical importance for providing adaptive scaffolding targeted to an individual student's needs. The goal of the research presented in this dissertation is to explore the use of a theoretically grounded machine learning framework for the real-time stealth assessment of SRL skills. In this framework, prominent psychological theories are used to guide the selection and application of machine learning and data mining techniques. Empirical models are trained on a corpus of data from students interacting with a game-based learning environment, CRYSTAL ISLAND. These models are evaluated to determine if the predictive models are benefited by the theoretically grounded framework and if they are also sufficiently robust to offer real-time trace-based assessment of SRL. The work consists of four phases of theory and data-driven investigation: (1) identification of SRL, (2) feature selection and creation, (3) model creation and (4) model evaluation. First, SRL skill measures of particular relevance to the CRYSTAL ISLAND environment are identified using a variety of techniques. Each measure is designed to be representative of one of the three components of SRL: cognition, metacognition and motivation. Students are classified based on their demonstration of these skills. Next, feature selection is used to identify simple personal attribute and in-game actions that are tied to SRL classifications. A differential sequence mining approach is then used to identify patterns of behavior that are indicative of each component of SRL. These patterns are used to create complex event features guided by SRL theory. In the model creation phase, these features are integrated in naive, static and dynamic Bayesian networks with structures that are increasingly informed by process models of SRL. The models are then empirically learned from a corpus of student interaction data. Finally, the models are evaluated to determine the value of the theoretically grounded empirical framework. Specifically, the evaluation seeks to quantify usefulness of the features identified in Phase 2 and the structures devised in Phase 3. Furthermore, the models are evaluated to identify how early into the interaction SRL skill can be accurately assessed and how well the models generalize to unseen populations. Overall, the results indicate that the theoretically-grounded machine learning framework is effective for developing real-time trace based assessment models of SRL. The differential sequence mining approach yields patterns of behavior that are both meaningful and improve the predictive accuracy of the models. Dynamic Bayesian models with temporal structures representing the cyclic nature of SRL outperformed those without this representation. Finally, the framework was effective at predicting the cognitive, metacognitive and motivational components of SRL within the first 20% of the interaction. These results highlight the potential of the framework for guiding the design of real-time trace-based SRL models.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Educational psychology. $3 555103
650 4 $a Educational technology. $3 556755
650 4 $a Computer science. $3 573171
653 $a Educational data mining
653 $a Game features
653 $a Game-based learning
653 $a Machine learning
653 $a Self-regulated learning
653 $a Self-regulatory skills
655 7 $a Electronic books. $2 local $3 554714
690 $a 0525
690 $a 0710
690 $a 0984
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a North Carolina State University. $3 845424
773 0 $t Dissertations Abstracts International $g 75-10B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3586200 $z click for full text (PQDT)