Knowledge extraction from Web user access logs

Recently, by rapidly developing and common usage of the Internet, Web has been largest accessible data source in the world. While extremely growing the knowledge masses up on the Internet by passing time, Web Mining has been seeing as more attractive subject more and more to cope with goals such as improving, growing healthy and using effectiveness of web site structure and to provide appropriate web service to the web clients requests.In this thesis, a new process intended for purifying of text-based web user access logs are proposed. The implementation and code of the proposed process have been designed on Java-based SAS software environment. Purifying high dimensional data access logs, the improved data cleaned process is superior to the other methods in speed aspect.Extracting meaningful and interesting knowledge from the purified user access logs, three different implementations have been realized.?Knowledge contained the meaningful and interesting patterns from web user access log files have been extracted by using Path Analysis Method. The implementation of web user access logs of Path Analysis Method, present in the literature and implemented in different fields, has successfully shown that it can be used in extracting meaningful and interesting knowledge.?Relation knowledge between web sites by applying Association Rules Method on the same dataset is extracted.?The detailed statistical knowledge regarding three months usage of the Web site has been extracted by using Statistical Analysis Method.In the end of all implementation, web site designers and managers are given suggestions about improvement, grow healthy and being usable of the web site and to contribute its structural organization by using obtained meaningful and pure knowledge. Furthermore, some solution suggestions with relation to increase successfulness of web site and server by analyzing HTTP state codes have been formed. The formed suggestions are intended for increase the visitor pleasure.