一种基于领域知识的链路预测方法  

A LINK PREDICTION METHOD BASED ON DOMAIN KNOWLEDGE

作　　者：李奋华 赵润林 Li Fenhua;Zhao Runlin(Maths&Information Technology School,Yuncheng University,Yuncheng 044000,Shanxi,China)

机构地区：[1]运城学院数学与信息技术学院,山西运城044000

出　　处：《计算机应用与软件》2020年第11期206-208,238,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(11501498)。

摘　　要：个性化推荐在现代的电子商务中获得了广泛的应用,但是传统的推荐算法存在冷启动、稀疏性和推荐精准度差等问题,并且没有考虑个体间存在的关联信息。针对上述问题,在电子商务中采用链路预测和领域知识相融合的方法进行推荐,并在相关数据集上进行实验研究。结果表明,该方法在电子商务推荐中能够有效地提高推荐的精准度,更贴近用户的需求,能取得较好的推荐效果。Personalized recommendation has been widely used in modern e-commerce,but the traditional recommendation algorithms have problems such as cold boot,sparsity and poor recommendation accuracy,and do not consider the related information among individuals in the process of recommendation.In order to solve the above problems,this paper uses the method of link prediction and domain knowledge to recommend in e-commerce,and carries out experimental research on related datasets.The results show that our method can effectively improve the accuracy of recommendation in e-commerce recommendation,more close to the needs of users,and achieve better results.

关 键 词：数据挖掘 社会计算 相似度 链路预测 个性化推荐 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]