The backdoor or information leak of Web servers can be detected by using Web Mining techniques on some abnormal Web log and Web application log data. The security of Web servers can be enhanced and the damage of illegal access can be avoided. Firstly, the system for discovering the patterns of information leakages in CGI scripts from Web log data was proposed. Secondly, those patterns for system administrators to modify their codes and enhance their Web site security were provided. The following aspects were described: one is to combine web application log with web log to extract more information,so web data mining could be used to mine web log for discovering the information that firewall and Information Detection System cannot find. Another approach is to propose an operation module of web site to enhance Web site security. In cluster server session, Density -Based Clustering technique is used to reduce resource cost and obtain better efficiency. The backdoor or information leak of Web servers can be detected by using Web Mining techniques on some abnormal Web log and Web application log data. The security of Web servers can be enhanced and the damage of illegal access can be avoided. discovering the patterns of information leakages in CGI scripts from Web log data was proposed. Secondly, those patterns for system administrators to modify their codes and enhance their Web site security were provided. These following aspects were described: one is to combine web application log with web log to extract more information, so web data mining could be used to mine web log for discovering the information that firewall and Information Detection System can not find. Another approach is to propose an operation module of web site to enhance Web site security. server session, Density-Based Clustering technique is used to reduce resource cost and obtain better efficiency.