基于项目传递打分和项目预测的协同过滤推荐

摘要

针对目前协同过滤中数据极端稀疏的问题,提出利用项目之间的传递打分来填充项目评分矩阵.在用户评分矩阵和项目矩阵的基础上使用项目预测的方法来降低稀疏性,提高推荐效率,并利用推荐算法产生推荐项目序列.最后,通过试验证明了该方法的可靠性和准确性.%To solve the problem of data extreme sparseness in the current cooperative filter system, firstly transferring scoring between items is proposed to fill project scoring matrix. Then the item prediction is employed to reduce the sparseness on the basis of item scoring and item matrix, which improves the efficiency of recommendation, and utilizes recommendation algorithm to produce item sequences. Lastly, the reliability and accuracy of this method is manifested through the trials.