一种有效缓解数据稀疏问题的协同过滤推荐算法

An efficient recommendation algorithm for relieving data sparsity in collaborative filtering

文章针对数据的稀疏会导致传统的协同过滤(collaborative filtering,CF)推荐算法不能准确地查找到最近邻居问题,提出了一种改进的基于用户Tanimoto相似性系数预填充的算法,通过改进的Tanimoto相似性系数得到更加合理的用户相似度,并结合提出的预测公式对目标用户的未评分项进行预测评分和填充,从而降低矩阵的数据稀疏度。实验结果表明,该算法对稀疏数据集具有较好的表现,能够提高推荐的质量。

Considering the data sparsity of user-item rating matrix, it is difficult to find the closest neighbors with traditional collaborative filtering(CF) recommendation algorithm. In this paper, a pre-filling algorithm based on improved Tanimoto similarity coefficient is proposed. The application of improved Tanimoto similarity coefficient is helpful for obtaining more reasonable user similarity, and a novel formula is used in the prediction of missing values, which can reduce the data sparsity of the matrix. The experimental results show that the proposed algorithm has better performance on sparse data sets and improves the quality of recommendation.