Applications of Web mining - from Web search engine to P2P filtering $

Abstract

We have developed Japanese Web search engine "Mondou (RCAAU)", which was based on the emerging technologies of data mining. Our search engine provides associative keywords which are tightly related to focusing Web pages. We also implemented the visual interface based on the technology of information visualization. In order to improve the performance of various search strategies by using characteristics of Web systems, we try to implement the advanced Web information systems with data mining and information technologies. Firstly, we introduce various Web mining algorithm, which efficiently reduces the computing cost of Web search. We pay attention to a part of useful pages effectively and improve the performance of Web search by using our proposed algorithms. Secondly, for preserving huge volume of born-digital information in the Internet, we are focusing on technologies of Web archiving system like WARP. In order to handle monotonously increasing digital information, we have to resolve many difficult problems of long life data preservation by improving Web searching techniques. Our experiences of our Mondou Web search engine and cooperative distributed Web robots are very useful and effective. Finally, the technologies of P2P (Peer-to-Peer) distributed search systems are becoming important rapidly. For example, it is very hard to discover appropriate information resources by simple queries of Gnutella, Freenet and so on. Therefore, in order to realize the topic-driven search, we propose more intelligent search systems, which are based on the technologies of data mining.