论文部分内容阅读
Slope One算法是一种基于内存的协同过滤推荐算法,在计算时,内存消耗过大,尤其当数据集稀疏时,预测结果的准确度不高.基于此,将一种基于模型的算法融合到基于内存的Slope One算法中,提出一种使用机器学习中最小二乘法改进的加权Slope One算法,该算法简单直观且计算高效,可以克服传统基于内存推荐算法的诸多缺点.最后,在Filmtrust和Movielens数据集上的对比实验结果表明,融合偏差因子的加权Slope One算法在这两个稀疏度不同的数据集下,均能获得较高的推荐准确度.
Slope One algorithm is a memory-based collaborative filtering recommendation algorithm, memory consumption is too large, especially when the data set is sparse, the accuracy of the prediction results is not high.Based on this, a model-based algorithm is integrated into In Slot One algorithm based on memory, a weighted Slope One algorithm based on least squares method in machine learning is proposed, which is simple, intuitive and computationally efficient, which can overcome many shortcomings of traditional memory-based algorithm.Finally, in Filmtrust and Movielens The experimental results on the dataset show that the weighted Slope One algorithm with fusion deviation factor can obtain higher recommendation accuracy under the two data sets with different sparsity.