[1]张 琪,武法提,许文静.多模态数据支持的学习投入评测:现状、启示与研究趋向[J].远程教育杂志,2020,(01):076-86.[doi:10.15881/j.cnki.cn33-1304/g4.2020.01.008]
 Zhang Qi,Wu Fati,Xu Wenjing.Learning Engagement Evaluation Supported by Multimodal Data: Status, Implications, and Research Trends[J].Distance Education Journal,2020,(01):076-86.[doi:10.15881/j.cnki.cn33-1304/g4.2020.01.008]
点击复制

多模态数据支持的学习投入评测:现状、启示与研究趋向()
分享到:

《远程教育杂志》[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2020年01期
页码:
076-86
栏目:
学习新论
出版日期:
2020-01-07

文章信息/Info

Title:
Learning Engagement Evaluation Supported by Multimodal Data: Status, Implications, and Research Trends
作者:
张 琪; 武法提; 许文静
1.淮北师范大学 教育学院,安徽淮北 235000;2.北京师范大学 教育技术学院,北京 100875
Author(s):
Zhang Qi; Wu Fati; Xu Wenjing
1.School of Education,Huaibei Normal University,Huaibei Anhui 235000;2.College of Education Technology,Beijing Normal University,Beijing 100875
关键词:
多模态数据建模学习投入智能评价研究趋向学习分析情感分析
Keywords:
Multimodality Data Modeling Learning Engagement Intelligent Evaluation Research TrendsLearning AnalyticsSentiment Analysis
分类号:
G420
DOI:
10.15881/j.cnki.cn33-1304/g4.2020.01.008
文献标志码:
A
摘要:
作为学习分析领域的重要内容,学习投入的评测日益成为研究热点。为此,阐释学习投入的概念与特征,指出经典学习分析的局限,即“路灯效应”(Streetlight Effect)有可能使研究者偏离解决“真实场景”中的问题,而多模态数据支持的学习评测契合了学习投入的动态、多维、境脉特征。多模态数据获取可以从交互情景中的行为分析、单模态传感器与多模态传感器三个维度来分类。多模态数据经过建模场景、数据源与精度等方面的刻画,可实现对学习者交互状态、辍学率、心智游移水平、注意力以成功表现等指标的评估,体现出对复杂认知能力衡量、改善建模精度以及对数据集整体意义还原的实践价值。未来学习投入的评测研究应强化对理论模型的构建,充分借助脑科学、教育神经科学的技术手段阐释学习者外部行为表现、认知过程与内部生理的相关机制,构建科学的生物数据库以及对脱离投入提供更为有效的解释与干预,从而为智能时代的个性化学习提供“增值”。
Abstract:
As the important content in learning analytics, the evaluation of learning engagement has become a research focus. To this end, the concepts and characteristics of learning engagement are explained, and the limitations of classical learning analytics are pointed out. That is, the "Streetlight Effect" may lead researchers to deviate from solving problems in "real scenarios", and the learning evaluation supported by multimodal data fits the dynamic, multi-dimensional, and contextual characteristics of learning engagement. Multi-modal data acquisition can be classified from three dimensions: behavior analysis in interactive scenarios, single-modal sensors and multi-modal sensors. Multi-modal data, characterized by modeling scenarios, data sources, and accuracy, can realize the assessment of learner interaction status, drop-out rate, mental migration level, attention and success performance, etc. Besides, it reflects practical value to complex cognitive abilities in measuring and improving modeling accuracy and restoring the overall meaning of the data set. The evaluation study of future learning engagement should strengthen the construction of theoretical models, fully utilizing the technical methods of brain science and educational neuroscience to explain the learners’ external behavior performance, cognitive processes and internal physiological related mechanisms. Meanwhile, it should build scientific biological databases and offer more effective explanations and interventions for engagement detachment, thus providing "added value" for personalized learning in the intelligent age.

相似文献/References:

[1]吴 忭,彭晓玲,胡艺龄.教育研究的去芜存菁之路:从多模态叙事到证据公平——美国AERA 2019年会述评[J].远程教育杂志,2019,(04):013.[doi:10.15881/j.cnki.cn33-1304/g4.2019.04.002]
 Wu Bian,Peng Xiaoling,Hu Yiling.Leveraging Educational Research in a “Post-truth”Era: From Multimodal Narratives to Democratize Evidence:Review of 2019 AERA Annual Meeting[J].Distance Education Journal,2019,(01):013.[doi:10.15881/j.cnki.cn33-1304/g4.2019.04.002]

备注/Memo

备注/Memo:
基金项目:本文系教育部人文社会科学青年基金项目“全息数据支持的学习投入建模与干预研究”(项目编号:18YJC880126)的研究成果。
更新日期/Last Update: 1900-01-01