Incremental web ranking on P2P networks

Web ranking is one of the most important components of web search services which becomes an important activity these days. In order to compute the web ranking, the web-link graph structure is to be processed to analyze the importance of the linkage. The time and space complexity for web ranking can be enormous as the number of web grows rapidly. Peer-to-peer (P2P) network computational models are an important approach to process such task efficiently. However, as mentioned that number of webs is increased continuously, a web ranking algorithm that considers the web-link graph as a static set of data may not be appropriated. When a snapshot of the web-link graph is being processed, the new change can occur. Thus, the ranking result can be inaccurate. In this paper, we proposed an efficient approach to incrementally compute web rankings on a P2P network. The proposed approach processes almost only the changed part of the web-link graph in the distributed manner, thus it performs the web ranking efficiently. Our experiment results show that the proposed approach can significantly reduce the computational cost as well as the communication cost.