基于用户属性和评分的协同过滤推荐算法

摘要

为解决协同过滤推荐系统数据稀疏和冷启动带来的问题，提出一种相似度计算和评分预测算法。结合用户评分相似度、兴趣倾向相似度和置信度3方面，更充分地利用用户评分信息，使得用户相似度的计算更准确、区分度更高；使用sigmoid函数，实现冷启动状态下用户相似度计算时用户属性和用户评分信息的平滑过渡。在MovieLens真实数据集上进行实验，实验结果表明，该算法可有效提高评分预测的准确性，在一定程度上解决冷启动的问题。%To solve the problem caused by data sparsity and cold start,a score predicting algorithm was presented.The score similarity,the interest tendency similarity and the confidence coefficient information were used to make the user similarity calcu-lation more accurate and discriminative.Besides,a smoothing method based on the Sigmoid function was adopted to make the transition between the user score and the user attributes information more smooth under the cold start state.Results of experi-ments on the MovieLens dataset demonstrate that the proposed algorithm can improve the accuracy of the score prediction,and solve the cold start problem to some extent.