基于社交关系与无监督学习的大数据推荐算法

针对大数据推荐系统中推荐准确率与效率较低的问题,设计一种基于社交关系与多上下文因素的大数据推荐系统。基于活动用户的社交网络,构建一个社交关系的张量模型;通过张量分解获得用户的上下文因素;基于候选集的相似性产生一个推荐列表。基于用户的反馈预测社交关系的范围,有效地减少推荐系统的计算量。真实数据集的实验结果证明,该算法提高了推荐系统的推荐精度,有效地缓解了稀疏性问题与冷启动问题,并且实现了较快的响应时间。

BIG DATA RECOMMENDATION ALGORITHM BASED ON SOCIAL RELATIONSHIP AND UNSUPERVISED LEARNING

To address the problem of the low accuracy and efficiency in big data recommendation systems,we proposed a big data recommendation system based on social relationship and multi-context factors.A tensor model of social relationship was constructed based on the social network of the active user.We obtained the contextual factors by tensor factorization.A recommendation list was generated based on the similarity of the candidate sets.The range of social relationship was predicted based on the feedback of users,so that the computational complexity of recommendation system was reduced effectively.Experimental results on the real dataset show that the proposed algorithm improves the accuracy of recommendation systems,effectively alleviates the sparsity and cold start problems,and achieves a faster response time.