FCAN:一种基于快速映射的内容访问网络  

FCAN: Fastmap-based Content Addressable Network

作　　者：王菁[1] 杨寿保[1] 高鹰[1] 郭磊涛[1] 

机构地区：[1]中国科学技术大学计算机科学与技术系,安徽合肥230026

出　　处：《系统仿真学报》2007年第17期3955-3960,共6页Journal of System Simulation

基　　金：国家自然科学基金项目(60273041);国家863重点项目(2002AA104560)

摘　　要：针对当前结构化P2P文件存储系统存在仅支持单关键字的精确匹配,而缺乏支持语义查询的局限性,提出一种基于快速映射的结构化P2P存储系统FCAN,利用FastMap实现从语义空间到P2P空间的映射,在保证文本矢量语义完整性的同时对其进行有效降维,从而实现基于内容的查找;通过对质心法确定的最近邻居进行聚类,有效地减少了冗余消息和查询时间。仿真实验表明FCAN不仅具有较高的召回率和精度,还具有较少的查询时间和冗余消息。The existing DHT-based file storage systems are suffering the problems of being restricted to exact match and lacking semantic-based query. To solve these problems, a DHT-based file storage system named FCAN was proposed. Mapping a document vector from semantic space to P2P space by FastMap, FCAN could effectively reduce the dimensions of it to implement semantic-based query without affecting the semantic integrality of this document. By clustering the nearest neighbors based on centroid, FCAN was improved to reduce the redundant messages and search time. Simulations show that FCAN has not only higher recall and precision, but also fewer search time and redundant messages.

关 键 词：FCAN FastMap 矢量空间模型 质心法 pSearch 语义小世界 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]