融合知识图谱关联检索的Web用户访问行为预测  

WEB User Access Behavior Prediction for Fusion of Knowledge Map Association Retrieval

作　　者：赵慧 ZHAO Hui(School of Information Science and Electrical Engineering,Shandong Jiao Tong University,Jinan Shandong,250357,China)

机构地区：[1]山东交通学院信息科学与电气工程学院,山东济南250357

出　　处：《计算机仿真》2023年第10期496-500,共5页Computer Simulation

摘　　要：挖掘网络用户在线行为规律、深入分析和预测访问行为信息能够提升网络安全、资源推荐精度。虽然用户访问存在一定规律性,但是随机性也是较为显著的特征,且考虑到访问数据的不完整性和错误概率,Web用户访问行为预测具有较大难度。为此,提出融合知识图谱关联检索的Web用户访问行为预测方法。挖掘存储于Web服务器中的用户访问行为数据,并剔除合集中不完整数据、错误数据和重复数据,提取用户访问行为数据特征。将数据特征与知识图谱结合,建立基于Web用户访问行为的融合知识图谱关键检索矩阵。利用DBSCAN聚类算法获取矩阵聚类结果,实现用户访问行为预测。实验结果表明,通过对比不同方法预测的节点链路与实际节点链路可知,所提方法的预测精确度更高,且耗时为0.97ms,说明该方法应用精度高、速度快。Mining online user behavior patterns,in-depth analysis and prefdiction of access behavior information can improve network security and resource recommendation accuracy.Although there is some regularity in user access,randomness is also a significant feature.Considering the incompleteness and error probability of access data,it is difficult to predict web user access behavior.To this end,a web user access behavior prediction method based on knowledge atlas and associated retrieval is proposed.Mining the user access behavior data stored in the web server,and removing the incomplete data,error data and duplicate data in the collection to extract the user access behavior data features.Combining data features with knowledge atlas,akey retrieval matrix of fused knowledge atlas based on Web user access behavior was established.The DBSCAN clustering algorithm was used to obtain the matrix clustering results and realize the prediction of user access behavior.The experimental results show that by comparing the node links predicted by different methods with the actual node links,the prediction accuracy of the proposed method is higher,and the time consumption is 0.97 MS,which indicates that the method has high application accuracy and speed.

关 键 词：用户访问行为 数据清洗 特征提取 知识图谱 聚类算法 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]