基于多维特征差异的个性化学习资源推荐方法

协同过滤算法和二进制粒子群算法是目前学习资源推荐领域研究热点.然而,协同过滤算法推荐的学习资源过于随机化,不能满足学习者进行整体知识建构的要求.而基于二进制粒子群算法构建的资源推荐模型,以推荐所有学习者完整的学习资源为目标,且模型数据较难预测,不符合在线智能化学习的趋势.针对以上问题,提出了基于多维特征差异的个性化学习资源推荐算法:首先根据学习者和学习资源多维特征差异建立学习资源推荐模型,并考虑了学习偏好;其次引入协同过滤技术对模型数据进行预测;最后针对推荐模型的多目标优化特征,将协同过滤算法和二进制粒子群算法结合,提出了对惯性权重和种群多样性进行动态协同调整的自适应二进制粒子群算法,实现了个性化学习资源推荐.实验证明,该算法具有较好的准确性,能够满足个性化学习资源推荐的需要.

The method of personalized learning materials recommendation based on multidimensional feature difference

Currently, in learning resources recommended field, researchers focus on collaborative filtering algorithm and binary particle swarm optimization (BPSO) algorithm. However, by using collaborative filtering algorithm, the learning resources are recommended with a too randomization way, which cannot meet the requirements of learners in building overall knowledge architecture. Furthermore, the recommended model based on BPSO algorithm asks to recommend the whole learning resources for all learners and the model data is hard to be predicted, which does not conform the development trend of intelligent online learning. In order to deal with the above problems, a personalized learning resources recommendation algorithm is proposed based on multidimensional feature differences. As a first step, learning resources recommended model is established according to the multidimensional feature differences in learners and learning resources, as well as the learning preferences. Next the collaborative filtering technology is adopted to predict model data. Finally, through combining the BPSO algorithm with collaborative filtering algorithm based on the multi-objective optimization characteristics of recommendation model, an adaptive binary particle swarm optimization algorithm is proposed to dynamically coordinate inertia weight and population diversity. As shown in the experiments, it is implemented that meeting the requirements in personalized learning resources recommendation with better precision.