A Study Of Information Sharing Agent Based On P2P

With the rapid development of networking technologies and the advent of peer-to-peer technologies, such distributed file-sharing systems as Napster and Gnutella, are becoming prevalent over time. Both systems possess their advantages as well as their disadvantages. Super-Peers networking overcome their disadvantages. We divide all peers into two groups: common peer and super peer. We look on the more powerful peer as super peer. These super peers will supervise the common peers near to it. Then,we introduced some search methods, and we focused more attention on a strategy which deal with the situation that both peer are not online simultaneously.As the information on the net is increasing dramatically, the accuracy of searching process of traditional method based on key words turned lower and lower. So we introduced metadata into the system. All the shared resource are divided into four types: papers, movies, songs and software. Each resource has his own metadata file of XML type. The search based on metadata proved more efficient and more accurate. We can attain the metadata manually and automatically. We can extract some important messages from the technology papers of doc type such as title, abstract, key words and writers. So we can get some important data automatically.From 1990s, the Agent technology has become one of the most important branch of AI(Artificial Intelligence) field. From the function view point, the Agent of each peer is an assistant. Each agent can provide with active service and intelligent search. We can learn the user's hobby by his search terms, so as soon as some new resource which fit for his hobby is added to our system, it will be informed the user. An ontology was built in our system which make the same meaning words be responding. If we enter a word, and we can find all the information related to this word.This thesis is organized as follows. Chapter 1 introduces background, previous work and related work. In chapter 2, we introduced the principal concepts of metadata and XML, then we described a new method on how to extract the important message from the doc type papers. In chapter 3, we introduced all kinds of P2P top structureand some traditional search methods at first, and then we put forward a new network structure: super-peer net structure. At last, we introduced some search methods, and described a search way which can handle with the situation where both side are offline. First, chapter 4 introduces the local information management, then we turned to the P2P technology. We focused two sides: active service and intelligent search. And conclusions are drawn in Chapter 5.