摘要
为充分利用显性特征和隐性特征的互补性,提出一种PtransE_CF视频推荐算法。在协同过滤中引入知识图谱推理技术,通过路径排序算法挖掘实体间多路径关系,将所有的实体关系嵌入到低维的语义空间中,在低维空间中计算任意视频间的语义相似性,将语义相似性与协同过滤的用户行为相似性结合进行推荐。实验结果表明,该方法弥补了协同过滤推荐算法对隐性信息利用不充分的缺陷,在语义层面增强了推荐的效果,在一定程度上解决了数据稀疏性问题。
To make full use of the complementarity of explicit features and implicit features,a PtransE_CF video recommendation algorithm was proposed.Knowledge graph reasoning technology was introduced to collaborative filtering,the path ranking algorithm was used to mine the multi-path relationship between entities,and all entities and relationships were embedded into the low-dimensional semantic space,the semantic similarity between randam videos was calculated.The semantic similarity was combined with user behavior similarity of collaborative filtering for recommendation.Experimental results show that the proposed method can make up for the deficiency that collaborative filtering algorithms cannot fully utilize the hidden information,and enhance the effectiveness of recommendation at the semantic level.What is more,to a certain extent,it can solve the problem of data sparseness.
作者
许智宏
赵杏
董永峰
闫文杰
XU Zhi-hong;ZHAO Xing;DONG Yong-feng;YAN Wen-jie(School of Artificial Intelligence,Hebei University of Technology,Tianjin 300401,China;Hebei Provincial Key Laboratory of Big Data Computing,Hebei University of Technology,Tianjin 300401,China)
出处
《计算机工程与设计》
北大核心
2020年第3期710-715,共6页
Computer Engineering and Design
基金
国家自然科学基金项目(61702157)
河北省科技支撑计划基金项目(15210506)
天津市自然科学基金项目(16JCQNJC00400、16JCYBJC15600)。
关键词
协同过滤
知识图谱
知识推理
语义相似性
路径权重
collaborative filtering
knowledge graph
knowledge reasoning
semantic similarity
path weight
