基于降维算法的分布式语义资源搜索

发布时间：2018-02-26 21:17

本文关键词： 向量模型坐标空间降维资源搜索对等网络　出处：《北京邮电大学学报》2013年02期 　论文类型：期刊论文

【摘要】：提出了一种面向高维资源的分布式相似资源搜索机制.针对传统的分布式对等(P2P)网络无法解决高维资源的相似性搜索问题,通过基于主成分分析的降维算法将高维资源向量模型映射到低维空间,以低维空间中资源向量模型为索引,映射到P2P网络里的分布式散列表中,以一种完全基于P2P网络和路由机制的简单有效方式实现分布式相似性资源搜索,同时避免资源维数过高引发搜索的维数灾难.对降维处理后资源相似性信息保留情况进行了分析,并通过基于内容寻址网络的仿真验证了降维算法对于构建低维资源索引的有效性.对于具有一定聚类特征的高维资源,该方法可以在分布式的相似性搜索中获得较高的查准率.
[Abstract]:We propose a new high dimensional resource distributed similar resources search mechanism. The traditional distributed peer-to-peer (P2P) network can not solve high dimensional resource similarity search problem, through the dimensionality reduction algorithm based on principal component analysis of the high dimensional resource vector model is mapped into a low dimensional space to a low dimensional space vector resources the model for the index, mapped to the distributed hash table P2P in the network, with a simple and effective way of P2P network and routing mechanism to implement distributed resources based on similarity search, while avoiding the curse of dimensionality of resources caused by too high. On the dimension of search dimension after resource similarity information retention were analyzed, and through the simulation based on content addressable network to verify the dimensionality reduction algorithm is effective for the construction of low dimensional resource index. For high dimensional resource has certain characteristic of clustering, the method can be divided in cloth High precision is obtained in the similarity search.

【作者单位】：北京邮电大学信息与通信工程学院;北京邮电大学网络技术研究院;
【基金】：杭州华星——北邮信通院2011研究生创新基金国家科技重大专项项目(2012ZX03005008)
【分类号】：TP393.02

【相似文献】