Web Log Preprocessing Method Based on Visiting Efficiency

Having analyzed the traditional way to process the visiting pattern of Web in detail; we come up with a new method based on users´ visiting efficiency of Web log to deal with the relative problems. The method applies two parameters, interest degree and familiarity degree, to research the Web user visiting patterns. Through studying the relationship of the two parameters and some steps we can resolve the coming problems which occupied at the moment of user sessions. Finally, we can carry it out to partition the user sessions according to the stable interest degree and accidental interest degree. In this way, we can attain the clean and effective data source for the Web data mining.