The stipulation of internet content rises dramatically in recent years. Servers have become extremely powerful and the bandwidth of end user connections and backbones grew constantly during the previous decade. Nonetheless, users frequently experience poor performance to access web sites or download files primarily if mobile devices have been used due to their limited storage, processing, display, power and communication resources. The causes are often performance which access directly on the servers (e.g. pitiable performance of server-side applications or during burst crowds) and network infrastructure (e.g. long geographical distances, network overloads,etc.). Hence, the goal of this study is to propose Rough Set (RS) as a knowledge representation for uncertainty in data of client behavior and mobile event specification with resource dependencies to reduce latency by prefetching selected resources to resolve the problems in handling dynamic web pages. We conducted the trace-based experiments on the RS approach for better classification outcomes.