摘要
预送作为主动cache,是cache机制由时间局部性向空间局部性的拓展.文章提出服务器主动预送的两种模式.基于单个URL的模式利用客户请求的马尔可夫链特征获取文档的时序相关模型,可进行多级预送.基于会话的模式包括基于文档属性和会话整体语义的途径,重点研究基于文档属性的途径,给出基本的聚集算法,探讨了文档兴趣的定量表达,提出反映访问时序的属性向量距离算法.对于预送性能的度量,给出请求命中率、会话命中率、预送效率和预送代价等度量方法,同时,完成大量实验,对客户行为分析的这两种模式进行比较.文章提出的由服务器访问记录提取客户行为模式的方法,不但适用于文档预送,对于服务器站点设计和ISP(internetserviceprovider)的服务规划也有重要价值.
Presending is an active service which extends caching mechanism from temporal locality to spatiallocality. Two modes of extracting user behavior patterns are proposed to predict future requests from clients forefficient presending. URL-based mode exploits the Markov-chain features of request series, and can be used forhierarchical presending. Session-based mode captures more semantics, and the authors' work emphasizes theclustering algorithm, feasible document weight definition, and attribute-vector-distance computation represent-ing order of accesses. Their performance is evaluated using appropriate metrics such as request hit rate, sessionhit rate, presending efficiency and presending cost. Numerous experiments are carried out to compare the twomodes. These methods are used for web presending, while they are helpful to web server design and ISP (inter-net service provider) service planning.
出处
《软件学报》
EI
CSCD
北大核心
1999年第11期1142-1147,共6页
Journal of Software
基金
国家863高科技项目
关键词
WWW
马中夫链
文档预送
WEB
WWW responsiveness, presending, request dependency, Markov chain, session clustering,document attribute, presending performance