摘要
跨境电商产品推荐由于受到语言和文化差异等原因,要实现精准推荐仅靠单一方法完全不够。为提高产品的有效推荐,采用混合式协同推荐策略,将隐语义挖掘和特征聚类算法联合应用于混合式系统推荐策略,并借助于Spark平台优化推荐效率。首先,采用隐语义模型(LFM)对用户及商品的隐含特征关注度和重要度进行初始化,并构建用户—商品评分函数;以RMSE为优化函数,通过梯度下降获得LFM用户—商品评分值,根据评分值生成候选商品推荐序列;接着采用K-means算法对用户—商品特征进行聚类分析,并通过鲸群优化算法(WOA)对初始类中心进行优化求解,获得候选商品推荐序列。综合两种策略得到商品推荐系列,生成最终用户推荐商品。仿真结果显示,通过Spark平台的LFM和WOA-K-means的混合式协同推荐,四家跨境电商平台均获得90%以上的商品推荐准确度,对大规模跨境电商产品具有较高的适用性。
Cross-border e-commerce product recommendation is subject to language and cultural differences and other reasons,to achieve accurate recommendation only a single method is not enough.In order to improve the effective recommendation of products,a hybrid collaborative recommendation strategy is adopted.Implicit meaning mining and feature clustering algorithms are applied to the hybrid system recommendation strategy,and the recommendation efficiency is optimized with the help of Spark platform.First,the LFM model was used to initialize the attention and importance of the hidden features of users and products,and user-product scoring function was constructed.RMSE was used as the optimization function to obtain the LFM user-product scoring value through gradient descent,and the candidate product recommendation sequence was generated according to the scoring value.Then K-means algorithm is used to cluster the user-product features,and WOA algorithm is used to optimize the initial class center,and the candidate product recommendation sequence is obtained.Synthesize the product recommendation series obtained from the two strategies to generate the end user recommended products.The simulation results show that through the hybrid collaborative recommendation of LFM and WOA-K-means of Spark platform,the four cross-border e-commerce platforms have obtained more than 90%product recommendation accuracy,which has high applicability for large-scale cross-border e-commerce products.
作者
李佳颖
刘静
LI Jia-ying;LIU Jing(School of Economics and Management,Guangzhou NanYang Polytechnic College,Guangzhou 510900,Guangdong,China;School of Computer Science and Technology,Kashgar University,Kashgar 844000,Xinjiang,China)
出处
《贵阳学院学报（自然科学版）》
2023年第4期38-43,共6页
Journal of Guiyang University：Natural Sciences
基金
2021年广东省特色新型智库“广东省乡村振兴高质量服务方略研究”(项目编号:2021TSZK021)
2022年广东省社科规划项目“‘数商兴农’工程下广东省农村电子商务竞争力评价研究”(项目编号:GD22XYJ28)。
关键词
跨境电商
商品推荐
Spark平台
LFM模型
Cross-border e-commerce
Commodity recommendation
Spark platform
LFM model
