摘要
Abstract
In order to deeply understand and mine the behavioral characteristics of user historical network interaction information,dynamically extract changes in user interaction behavior,and achieve personalized recommendation of network information,a knowledge graph based network information self-supervised reinforcement learning recommendation model is constructed.In the model,a knowledge graph of user network information interaction behavior is constructed to clearly display user′s historical network information interaction behavior.The dynamic changes of user behavior in the knowledge graph can be captured effectively by means of the feature extraction model based on self-supervised reinforcement learning to avoid the negative impact of popularity bias,so as to extract the features of historical network interaction information.Based on knowledge graph similarity calculation,the network information entities with similar features to user historical interaction information are recommended to realize the accurate and personalized recommendations.The experimental results verifies that after recommending online movie information resources to users,the click play conversion rate can reach 96.83%,and the personalized recommendation effect of online information is improved significantly.关键词
知识图谱/网络信息/自监督/强化学习/推荐模型/交互信息/特征提取/相似度计算Key words
knowledge graph/network information/self supervision/reinforcement learning/recommendation model/interactive information/feature extraction/similarity calculation分类
信息技术与安全科学
