Abstract: Mining of association rules is an important research topic in web usage mining Currently, web log association rules mining algorithms neglect an important problem of whether users are interested in the rules or not web log association rules mining algorithms are studied Combined with web topology structure, a web topology probability model and an interesting association rules （IAR） algorithm are presented Using web topology probability model to evaluate association rules’ interest, IAR gains high interest rules, which can be used to improve network performance The experiment shows how IAR enhances rules’utilization and effectively improves web topology It can be successfully applied to web usage mining