国外近十年深度学习实证研究综述
2019-06-14沈霞娟张宝辉曾宁
沈霞娟 张宝辉 曾宁



[摘 要] 为厘清国外教育领域深度学习的研究进展,选取Web of Science、Springer、ScienceDirect等数据库的85篇文献(2008—2018年),依据实证研究的核心要素建立编码体系进行内容分析。结果表明:(1)研究主题集中在深度学习策略、方式和评价三个方面;(2)比格斯、马顿&塞尔乔、恩特威斯尔&拉姆斯顿提出的深度学习概念框架得到较多认可;(3)深度学习在各类学科情境中广泛应用,但研究对象多为大学生;(4)实证研究数据呈现以量化分析为主、质性分析为辅的特征,混合分析日益得到重视;(5)自我报告量表、编码标准、条件化测量、眼动追踪是深度学习的主要测量方法;(6)深度学习的有效性突出表现在学业成绩、知识理解、学习体验、思维与能力发展等方面。
[关键词] 深度学习; 实证研究; 文献综述; 研究主题; 研究方法; 研究情境
[中图分类号] G434 [文獻标志码] A
[作者简介] 沈霞娟(1982—),女,河南商丘人。讲师,博士研究生,主要从事学习科学、教育信息资源等方面的研究。E-mail:shenxiajuan2009@163.com。张宝辉为通讯作者,E-mail:baohui.zhang@snnu.edu.cn。
[Abstract] In order to clarify the research progress of deep learning in education abroad , 85 papers are selected from Web of Science, Springer, ScienceDirect and other databases (2008-2018). Then a coding system is established based on the core elements of empirical research for content analysis.The results indicate that: (1) the research themes mainly focus on three aspects of deep learning strategies, approaches and evaluation; (2) The conceptual frameworks of deep learning proposed by Biggs, Marton & Saljo and Entwistle are widely recognized; (3) Deep learning is closely used in various subject contexts, but the research objects are mostly college students; (4) Research data is mainly analyzed by quantitative methods , which is supplemented by qualitative methods, while mixed analysis is increasingly valued; (5) Self-report scale, coding standard, conditional measurement and eye movement tracking are the main measuring methods of deep learning; (6) The effectiveness of deep learning is highlighted in academic performance, knowledge understanding, learning experience, the development of thinking and competence.
[Keywords] Deep Learning; Empirical Study; Literature Review; Research Themes; Research Method; Research Context
一、研究背景
面对知识更迭速度的加快,以及经济全球化、信息化带来的发展挑战,人们开始对学生的学习方式和质量进行深入反思。在基础教育领域,世界各国纷纷通过颁布核心素养的方式提升学习目标。观察这些核心素养的内容与结构,不难发现,为了成为合格的21世纪公民,学习者不仅要牢固掌握学科知识,还要学会学习,具有良好的问题解决、高阶思维、自主学习和实践能力。因此,机械的、被动的、简单记忆式的浅层学习已经无法满足社会发展的需要。……
