Abstract: In order to solve problem of difficult finding information for users of coal mine enterprise Web site, the paper put forward a personalized recommendation service model of coal enterprise Web site based on Web log. The model uses association rules to recommend page for new users, and uses clustering algorithm to recommend page for old users. And it measures interesting degree of users with click number of Web page, browsing time of Web page, Jacobian coefficient and the longest public path coefficient of a user to accurately recommend interested pages for the user. The test results showed that the model can classify Web resources and provide personalized recommendation.