基于LQN模型的租户兴趣浏览路径挖掘  

Mining interested browsing path of tenants based on LQN model

作　　者：谢琪琦 陈燕[1] 陈宁江[1] 李湘[1] 梁小宇[1] 

机构地区：[1]广西大学计算机与电子信息学院,广西南宁530004

出　　处：《重庆邮电大学学报（自然科学版）》2014年第6期756-762,共7页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：国家自然科学基金(61063012;61363003);广西自然科学基金(2012GXNSFAA053222);广西高校优秀人才资助计划([2011]40);广西科学研究与技术开发计划(桂科软13180015)~~

摘　　要：面向云计算环境中多租户应用的租户个性化服务需求,从多租户应用日志记录挖掘出租户兴趣浏览路径受到关注。针对传统的以浏览频度为主体的网络拓扑图研究问题,为了更好地找出租户在云平台网站上的兴趣网页,挖掘租户的兴趣浏览路径,综合租户对网页的浏览时长、接收字节数和浏览频度等多个要素定义租户兴趣度,构造租户执行图,纠正路径交叉状况以消除租户执行图中存在的多余路径,对循环路径中的对等节点进行整合以消除循环嵌套,给出了一种基于分层排队网(layered queue network,LQN)模型的租户兴趣浏览路径挖掘方法,在此基础上,借助广度优先遍历(breadth first search,BFS)算法进行挖掘。实验证明,改造后的LQN模型在租户兴趣浏览路径挖掘方面的效率有所提高。To meet the tenants＇ personalized service demand of multi-tenant applications in cloud computing environment, mining interested browsing path of tenants from multi-tenant application logs has become a concerned problem. The tradi- tional research of network topology is based on browsing frequency. In order to find out the interested web pages on a cloud platform and dig the interested browsing path of tenants, the paper integrates browsing time, receiving bytes and browsing frequency together to define the tenants＇ interest. To construct the tenant execution graph, paths crosses are corrected in or- der to eliminate the redundant paths and peer nodes are integrated for eliminating the loop nesting in cyclic paths, so that a method of mining tenants＇ interested browsing path which is based on layered queue network （ LQN ） model is proposed. Based on the presented method, breadth first search（BFS） method is used to mining interested browsing path. Experimen- tal results show that the improved LQN model is more effective than the traditional network topology in mining interested browsing path of tenants.

关 键 词：分层排队网(LQN)模型 兴趣浏览路径 兴趣网页 多租户 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]