摘要
传统的推荐算法随着用户和项目的数量增多,新用户在单一项目上的行为减少,导致推荐质量较低,鉴于此,提出一种融合文档主题算法(LDA)和交替最小二乘算法(ALS)的混合协同过滤推荐算法。LDA-ALS算法结合了文档主题算法和交替最小二乘算法的优势,缓解因用户信息缺失造成的冷启动问题,并将高维的用户-项目评分矩阵映射到低维的近似矩阵中,有效缓解了数据稀疏性问题。实验结果表明:在Spark平台下,该算法在旅游数据集上比传统推荐算法降低了2.4%的误差,而且更能适应目前网络环境下的大数据处理。
With the increase in the number of users and items,the traditional recommendation algorithm reduces user behavior on a single item,and the data of new users and items added is insufficient,resulting in lower recommendation quality.In view of this,a hybrid collaborative filtering recommendation algorithm composed of document topic algorithm(LDA)and Alternating Least Squares(ALS)was proposed.The LDA-ALS algorithm combined with the advantages of the document topic algorithm and the alternating least squares algorithm to alleviate the cold start problem caused by the lack of user information,and the high-dimensional user-item rating matrix was mapped to the low-dimensional approximate matrix,the problem of data sparsity was effectively alleviated.Experimental results show that under the Spark platform,the algorithm reduces the error of 2.4%compared with the traditional recommendation algorithm on the tourism data set,and it is more adaptable to big data processing in the current network environment.
作者
陈丽芳
陈宏松
孙海民
CHEN Li-fang;CHEN Hong-song;SUN Hai-min(College of Science,North China University of Science and Technology,Tangshan Hebei 063210,China;The Technology Innovation Center of Cultural Tourism Big Data of Hebei Province,ChengdeHebei 067000,China;Hebei Normal University for Nationalities,Chengde Hebei 067000,China)
出处
《华北理工大学学报（自然科学版）》
CAS
2022年第1期89-97,共9页
Journal of North China University of Science and Technology：Natural Science Edition
基金
河北省文化旅游大数据技术创新中心开放课题(SG2019036-yb2005)。
