一种基于特征提取的教育视频资源推送方法

The Implementation of Educational Video Resources Recommendation Method Based on Feature Extraction

丰富的网络教育视频资源满足了学习者自主选择学习内容、时间和地点的需求。然而资源自身及其平台存在着内容质量参差不齐、优质教育资源匮乏、资源同质化、资源推送方式单一等问题,学习者难以快速高效地从海量的资源中获取与自身需求相关的有价值的资源。为让学习者高效获取个性化教育视频资源,在进行资源推送时,研究采用深度学习方法准确识别出视频资源中的知识点,通过构造视频中的知识点、视频质量和学习者需求之间的特征向量作为支持向量机的输入,由支持向量机决定是否将视频资源推送给学习者。将学习者对推送结果的实际点击率和学习者反馈的满意度作为方法的性能评价指标。这种推送方法关注学习者的兴趣需求和视频特征的结合,能更好地满足学习者的要求并提升学习效率,具有较大的应用潜力。

The plentiful network educational video resources meet the demand of learners to select learning content, learning time and learning place by themselves. However, it＇s difficult for learners to obtain worthwhile resources associated with their needs from the massive resources quickly and efficiently because of some problems existing in the resources or the resources platform such as the varied quality of resource content, the scarcity of high-quality educational resources, resources homogenization, the single mode of resources pushing. To allow learners to get personalized educational video resources efficiently, when recommending resources, this research uses deep learning method to identify the knowledge points of video resources accurately. Then the feature vector structured by the knowledge points of video resources, video quality and the needs of learners serves as the input of the support vector machine, which is responsible for deciding whether to recommend video resources to the learners or not. The performance evaluating indicators of this proposed method includes the actual click rate of the recommended resources by learners and the satisfaction degree fed back by learners. This proposed method focuses on the combination of the learners＇ interests and video features, which can better meet the requirements of learners and enhance the ability of learners with great potential in application.