VLE

VLE数据集是一个包含约12,000个科学视频讲座的新型数据集，由伦敦大学学院人工智能中心创建。该数据集涵盖了计算机科学、人工智能和数据科学等领域的讲座，提供了丰富的文本和视频特定特征，以及与学习者参与度相关的隐式和显式信号。数据集的创建旨在解决大规模学习资源集合中的冷启动问题，特别是在电子学习和MOOC环境中。通过使用VLE数据集，研究人员能够构建和验证无上下文参与度预测模型，这些模型在性能上显著优于基于先前数据集的模型。此外，数据集还支持多种任务，如质量保证和教育推荐系统的改进，特别是在处理新用户或新内容时的冷启动问题。

The VLE dataset is a novel dataset consisting of approximately 12,000 scientific video lectures, created by the Artificial Intelligence Centre at University College London. This dataset covers lectures across computer science, artificial intelligence, and data science, and provides abundant text and video-specific features, as well as implicit and explicit signals related to learner engagement. The dataset was developed to address the cold start problem in large-scale learning resource collections, particularly in e-learning and MOOC environments. By using the VLE dataset, researchers can construct and validate context-free engagement prediction models, which significantly outperform models based on prior datasets. Furthermore, the dataset supports multiple tasks such as quality assurance and the improvement of educational recommendation systems, especially in resolving cold start issues when dealing with new users or new content.