A Comparison of Prediction Algorithms for Prefetching in the Current Web

Abstract

[EN] This paper reviews a representative subset of the prediction algorithms used for Web prefetching classifying them according to the information gathered. Then, the DDG algorithm is described. The main novelty of this algorithm lies in the fact that, unlike previous algorithms, it creates a prediction model according to the structure of the current web. To this end, the algorithm distinguishes between container objects and embedded objects. Its performance is compared against important existing algorithms, and results show that, for the same amount of extra requests to the server, DDG always outperforms those algorithms by reducing the perceived latency up to 70% more without increasing the complexity order.