P2P网络中相似文本的聚集存储和模糊查询

针对目前各种结构化P2P协议在对资源进行查询时依赖于准确的全局唯一标识符(Global Unique Identifier),无法实现相似资源的模糊查找的问题,提出了一种资源发布和查找策略,实现了相似文本的聚集存储和模糊查询。该策略首先根据文本内容生成具有强鲁棒性的指纹,以该指纹作为文本在P2P网络中发布的基础,实现将相似的文本发布到同一节点或相邻节点,进而实现高效的分布式相似文本查询。该算法构建在分布式哈希表(DHT)路由基础之上,适用于各种以DHT为特征的结构化P2P系统。

Aggregating store and approximate text addressing in P2P networks

To launch a search for a given resource in different existing structured Peer-to-Peer(P2P) networks,the global unique identifier of the resource must be known in advance.Hence it is difficult to implement a search for approximate documents.To solve this problem,a novel routing scheme was proposed for resource publishing and addressing in structured P2P systems.By this scheme,according to the content of the texts a robust fingerprint is generated,which is taken as the base of text publication in P2P network.The key-value pairs of approximate tests are published to a unique node or contiguous peers,thus the distributed approximate texts can be efficiently searched by range queries.The system is constructed based on the Chord overlays,it is also applicable on other structured P2P systems.