融合知识图谱表示学习的栈式自编码器推荐算法  

RECOMMENDATION ALGORITHM BASED ON REPRESENTATION LEARNING OF KNOWLEDGE GRAPH AND STACK AUTOENCODER

作　　者：王卫红[1] 冯倩 吕红燕 曹玉辉[1] Wang Weihong;Feng Qian;LüHongyan;Cao Yuhui(School of Information Technology,Hebei University of Economics and Business,Shijiazhuang 050061,Hebei,China)

机构地区：[1]河北经贸大学信息技术学院,河北石家庄050061

出　　处：《计算机应用与软件》2021年第2期264-269,共6页Computer Applications and Software

基　　金：留学回国人员择优资助项目(C2015003042);河北省自然科学基金青年项目(F2015207009)。

摘　　要：针对目前协同过滤推荐算法中数据稀疏和语义信息欠缺问题,提出一种融合知识图谱表示学习的栈式自编码器推荐算法(SAEKG-CF)。将评分矩阵作为栈式自编码器的输入,训练得到项目的隐性特征向量,并据此计算特征相似性矩阵;利用知识图谱表示学习算法将项目中的实体映射到低维向量空间,并计算出低维向量空间中实体间的语义相似性矩阵;将特征相似性矩阵与语义相似性矩阵相融合,得到融合相似性矩阵,进而依据最优融合相似性矩阵产生top-k推荐列表。实验结果表明,该算法能有效地同时解决数据稀疏与语义信息欠缺问题,提高推荐的准确率。Aiming at the problem of data sparseness and lack of semantic information in the current collaborative filtering recommendation algorithm,a stacked autoencoder recommendation algorithm based on knowledge graph representation learning(SAEKG-CF)is proposed.It used the rating matrix as the input of stack self-encoder,trained the implicit feature vector of the project,and then calculated the feature similarity matrix;the knowledge graph representation learning algorithm was used to map the entities in the project to the low-dimensional vector space,and calculated semantic similarity matrix between entities;the feature similarity matrix was merged with the semantic similarity matrix to obtain a fusion similarity matrix;according to the optimal fusion similarity matrix,a top-k recommendation list was generated.The experimental results show that the proposed algorithm can effectively solve the problem of sparse data and lack of semantic information,and improve the accuracy of recommendation.

关 键 词：协同过滤 栈式自编码器 知识图谱 推荐系统 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]