融合知识图谱的学习者个性化学习资源推荐

Learner Personalized Resource Recommendation Based on Knowledge Graph

针对学习资源个性化推荐中存在的数据稀疏和冷启动问题,本文提出了一种基于知识图谱的学习者个性化学习资源推荐模型模型(LPRM).LPRM模型利用在线学习中学生与课程的历史交互信息以及在线课程的属性信息构建课程知识图谱,辅助课程资源的个性化推荐;针对RippleNet框架中实体关系传播未考虑实体影响力的问题,提出节点影响力计算模型衡量知识图谱中实体的影响力,构建LPRM模型框架得到学习者对学习资源的评分.大量对比实验结果表明,本文提出的基于知识图谱的学习者个性化学习资源推荐模型模在AUC和ACC评价指标上均表现出最优的性能,模型参数分析结果表明LPRM模型能有效地提升学习者学习资源个性化推荐性能,较好地缓解了数据稀疏和冷启动引起的学习者个性化资源推荐不准确等问题.

Aiming at the problem of data sparsity and cold start in personalized recommendation of learning resources,this paper proposes a learner personalized learning resource recommendation model model based on knowledge graph(LPRM).LPRM model uses the historical interaction information between students and courses in online learning and the attribute information of online courses to construct the course knowledge map and assist the personalized recommendation of course resources.Aiming at the problem that entity influence is not considered in the propagation of entity relationship in RippleNet framework,a node influence calculation model is proposed to measure the influence of entities in knowledge graph,and the LPRM model framework is constructed to obtain the scores of learners on learning resources.A large number of comparative experimental results show that the entity weighting model based on knowledge graph presented in this paper shows the optimal performance in both AUC and ACC evaluation indexes.The model parameter analysis results show that LPRM model can effectively improve the performance of personalized recommendation of learning resources.It can better alleviate the inaccuracy of personalized resource recommendation caused by data sparsity and cold start.