Since web is increasingly important for individuals, analyzing user behaviors on webpages is becoming essential for system designers and website providers. But current solutions cannot meet high demands on high accuracy, low time consuming and sense of privacy. Therefore, this paper designed a fuzzy clustering algorithm based web interaction system, presented multiple linear regressions to analyses user’s browsing behavior. Finally, we designed a couple of experiment to demonstrate the efficacy of this paper design system. The result shows the interaction system worked with high accuracy, low time consuming and good performance of scalability.