一种基于贝叶斯分类的个性化导购推荐算法

摘要

顺应电子商务的发展趋势,结合导购网站的个性化内容结构,提出了基于协同过滤的推荐算法与贝叶斯分类算法相结合的混合推荐算法.两种经典算法通过对用户评分值加权相加进行混合,从而得到用户对物品最终的预测评分,通过预测评分的高低进行推荐.通过这种混合推荐算法,可以更加充分地利用个性化导购网站的特殊结构化内容结构,从而达到提高个性化导购网站推荐系统的推荐内容准确度的目的.实验证明,改进后的混合算法在相同的条件下能够获得更低的平均绝对误差(MAE)和更好的准确率及召回率.%Following the trend of e-commerce and considering personalized content structure of shopping guide website,a new hybrid recommendation algorithm which is based on classic collaborative recommendations and content-based recommendations is proposed.Simultaneously,Bayesian classification is used in content-based recommendation algorithm.To get the final score of the hybrid algorithm,scores of collaborative and content-based algorithm are combined with particular weight numbers.After getting final scores,recommended items can be listed.Through this hybrid recommendation algorithm,it can take advantage of special content structure of personalized shopping guide website more efficiently so that we can improve the accuracy of the recommended content pushed by recommend system.The experimental results show that under the same conditions,it can reduce mean absolute error and improve precise and recall with the hybrid recommendation algorithm.