Web Prefetching With Machine Learning Algorithms

Prefetching algorithms can effectively reduce web latency and dramatically improve responsiveness of interactive web applications. The authors propose a new history-based prefetching algorithm assisted by data mining techniques that achieves very high prediction accuracy, generates little overhead traffic, and allow users to control the amount of memory that it uses. The authors also propose a method to find accurate upper bounds on the performance of any history-based prefetching algorithm. Even with the speed of today's internet, web latency is still one of the major concerns of its users.