| 王觅, 陈道杰, 展金梅.深度学习技术在智能学习评价中的应用[J].海南师范大学学报自科版,2024,37(4):522-529 |
| 深度学习技术在智能学习评价中的应用 |
| Applications of Deep Learning Technology in Intelligent Learning Evaluation |
| |
| DOI：10.12051/j.issn.1674-4942.2024.04.016 |
| 中文关键词: 深度学习 智能学习评价 行为识别 多模态数据 |
| 英文关键词: deep learning intelligent learning evaluation behavior recognition multimodal data |
| 基金项目:海南省自然科学基金项目（721QN0890）；海南省高等学校教育教学改革重点项目（Hnjg2021ZD-20）；海南省高等学校教育教学改革研究项目（Hnjgzc2023-73）；海南省自然科学基金高层次人才项目（623RC481） |
|
| 摘要点击次数: 563 |
| 全文下载次数: 148 |
| 中文摘要: |
| 随着新一轮科技革命在教育评价领域的加速渗透，深度学习作为人工智能的重要研究方向之一，目前在智能学习评价领域展现出了巨大的应用潜力。利用深度学习技术构建复杂的神经网络模型，自动提取学习过程中的多维数据特征，能够实现对学生学习状态、学习进度和学习成效等的精准识别与评价。同时，针对不同学生的学习需求，深度学习技术能够提供个性化的学习建议和资源推送，进而促进学生自主学习和全面发展。因此，基于深度学习技术的智能学习评价不仅对我国教育评价改革具有重要意义，对于推动教育数字化、智能化转型也具有重要的理论和实践价值。为促进深度学习在智能学习评价领域的进一步应用，本文运用文献学研究方法，通过人工过滤方式对中国知网（CNKI）1423篇相关研究文献进行筛选统计分析，剔除与主题明显无关以及非技术支持的深度学习（即教育领域中的深度学习）相关文献后得到33篇研究样本，以此归纳出当前深度学习技术在智能学习评价中应用的3个主要研究方向：学习行为和情感特征的识别与分析，多模态数据的采集、分类与融合，个性化资源的推送与服务。进一步分析了目前存在的问题及挑战，包括评价结果可解释性较差、模型泛化能力不足以及数据更新与保密问题等，提出了完善智能学习评价模型、构建人机协同的评价体系、协同构建大规模共享数据库、技术融合促进数据安全4个发展建议，以期为后续相关研究提供参考和借鉴。 |
| 英文摘要: |
| With the accelerated penetration of an new round of technological revolution in the field of educational evaluation, deep learning, as one of the significant research directions of artificial intelligence, has shown great application potential in the field of intelligent learning evaluation. By leveraging deep learning technology to construct complex neural network models and automatically extract multidimensional data features from the learning process, it is possible to accurately identify and evaluate students' learning states, progress, and outcomes. Furthermore, deep learning technology can provide personalized learning suggestions and resource pushes tailored to the diverse needs of students, thereby promoting students' autonomous learning and comprehensive development. Consequently, intelligent learning evaluation based on deep learning technology has not only substantial significance for China's educational evaluation reform, but also important theoretical and practical value for promoting the digital and intelligent transformation of education. To facilitate the further application of deep learning in the realm of intelligent learning evaluation, this paper used the method of literature research to screen and statistically analyze 1,423 relevant research literatures on CNKI through manual filtering. After excluding literature that is evidently unrelated to the topic and non-technical support for deep learning (i.e., deep learning within the educational domain), a final sample of 33 research articles was obtained. Based on this, the paper summarized three main research directions for the application of current deep learning technology in intelligent learning evaluation including the recognition and analysis of learning behavior and emotional characteristics, the collection, classification, and fusion of multimodal data, and personalized resource push and services. Further analysis of the current problems and challenges, including poor interpretability of evaluation results, insufficient generalization ability of models, and data update and confidentiality issues are conducted. Four development suggestions, including improving intelligent learning evaluation models, establishing a human-machine collaborative evaluation system, collaboratively building a large-scale shared database, and promoting data security through technological integration are proposed, aiming to provide references and insights for subsequent related research. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
