Abstract: Similar customer groups, relevant Web pages, and frequent access paths can be discovered by analyzing of Web log files and customer database. In this paper, novel Web log mining algorithms are presented. First, according to Web site’s directed graph defined, a URL UserID relevant matrix is set up, where URL is taken as row and UserID is taken as column, and each element’s value of this matrix is the user’s hits. Second, similar customer groups are discovered by measuring similarity between column vectors, and relevant Web pages are obtained by measuring similarity between row vectors; frequent access paths can also be discovered by further processing of the latter. Experiments show the effectiveness of the algorithms.