Predictive web prefetching using mouse movement

Abstract

The delay of web page loading time becomes one important factor of user experience. Lots of users are impatient. Therefore reducing the delay is important for both individuals and companies. This paper will use the prefetching techniques to predict and fetch the next clicked links objects before the user clicks on that link to reduce user-perceived latency. Though lots of prefetching techniques are already studied, none of them use user mouse movement trace to do prediction. This paper will deploy the trace as source and will examine three simple heuristics for the prediction. Each heuristic will be evaluated through simulation and implemented. The results show that they would work well on average under certain parameter values and there are still limitations to be improved.