Research And Realization Of P2P File-searching Mechanism Based On JXTA

The primary reason for using P2P network technology is to realize the computing power between autonomic organizations and the sharing in the data resources. In consideration of a large number of data in P2P network existing in the form of file, one of the most difficult problems to the technology is how to describe and organize the files in the network so that users and programs can access it conveniently. To solve this, people get down to the research of P2P NFS. After referring to various documents, we study the JXTA virtual network and file search strategy in detail and in depth, and in combination of the protocols of JXTA and the properties of practical systems, the main job of the paper includes as follows:Firstly, study the available JXTA search mechanism and then analyze its limitations. Since there are no restrictions to peer group in the protocol of JXAT, it's hard for us to decide how much influence the common concept of peer group has to the efficiency of discovering resources in JXAT, so the concept of peer group has no prominence in this mechanism. Meanwhile, the available JXTA network resources discovery focus on the way of DHT which has unconsolidated consistency. Though linking it to rambler in limited range, it's truly a kind of the DHT network after all. Due to no support to search on the basis of content in the DHT network, this kind of search system limits its use in search and commonality.Secondly, by integrating with practice, the paper proposes a classified and complicated search model GCCS which connects with the concept of peer group. In our system, there are the same tasks and interest in essence to all peers in the same group, so the concept of peer group is positive to search performance. The model of GCCS integrates with the concept of peer group well and adopts the search and propagation mode which is cascaded in peer group. When users release their sharing, means of classification based on VSM have been applied to the auto-classification of the sharing resources. So when other users have more complicated search requirements to search matching, we compute the distance of vectors in the local related class files to find out the required files, according to the submitted keywords and related category.Thirdly, GCCS model has been applied to shared-file system which enhances the search performance of the system. Our system provides a convenient and quick function to deliver and acquire files. With GCCS model applied to the file search module, our system not only supports accurate search in the group and the whole networks, but also realizes complicated search in the networks so that it can satisfies users, enhances the flexibility of shared-file system and improves the efficiency of the cooperative work.