【作 者】：
杨现民， 袁 萌， 李康康
【关 键 词 】：
网络学习资源； 学习资源进化； 预警模型； 深度学习； 卷积神经网络
【栏 目】：
学习环境与资源
【中文摘要】：
大规模开放协同环境给网络学习资源的高效管理带来了新挑战，基于浏览量与评分排名的传统管理方法难以保证高质量网络学习资源的生成与持续进化。如何建构智能化的网络学习资源进化预警模型，是大数据时代破解网络学习资源高效管理的核心问题。该研究以网络学习资源进化预警任务的实施流程为依据，在探讨资源进化态表征、资源进化要素项提取、资源进化态标注、资源进化预警任务的数学定义等问题的基础上，构建了网络学习资源进化预警的技术模型。通过学习元平台的两个数据集实验验证表明，研究提出的预警模型在预测精准率、F1和AUC等指标方面表现优于对比算法，能较好地完成网络学习资源进化预警的任务。
【英文摘要】：
Large-scale open collaborative environment has brought new challenges to the efficient management of e-learning resources. The traditional management methods based on page views and rating rankings are difficult to guarantee the generation and continuous evolution for high-quality e-learning resources. How to construct an intelligent early warning model for the evolution of e-learning resources is the core problem of solving the efficient management of e-learning resources in the era of big data. Based on the implementation process of e-learning resource evolution warning task, this study constructs a technical model of e-learning resource evolution warning on the basis of exploring the issues of resource evolution state representation, resource evolution element item extraction, resource evolution state labeling, and mathematical definition of resource evolution warning task. The experimental performance on two datasets extracted from the Learning Cell Platform shows that the proposed early warning model outperforms the comparison algorithm in terms of predicting accuracy, F1, AUC and other indicators.