Research on Information Retrieval Based on Keywords Relationship and Past Retrieval Records in P2P Networks

Due to the inability of traditional keyword-based retrieval using semantic meaning in Peer-to-Peer networks, large bandwidth is wasted; meanwhile desirable information is still far beyond the search results most algorithms could gain. How to locate the needed peer accurately and how to obtain requested information rapidly are important issues. In this paper, through an automatic study mechanism toward a series of keywords inputted by user´s operation and past retrieval information tracking, an algorithm used to dynamically build the keyword semantic relationship table and to generate the index table of the keywords and their corresponding peers is proposed. The utilization of these tables promotes the accuracy in locating the needed peers and the efficiency in retrieving information. By assistance of feedback mechanism, knowledge is advanced to share within the interest domain. We also present the experiment and results prove efficiency and low-cost of the algorithm.