Predicting End-Users' Behaviors by Applying Microprocessor Branch Prediction Algorithms to Session Logs of a Networked Community

摘要

If people can predict the future, they can organize their schedules more efficiently and complete their work in a more timely fashion. Many software applications try to assist users by utilizing some kind of prediction. For example, knowledge management systems provide various personalized recommendation tools that analyze each end-user's past behaviors and, based upon this analysis, predict their next actions. In the research domain for computer architecture, many architects have tried to develop accurate prediction algorithms in order to increase the performance of microprocessors. Adopting microprocessor branch predictions for the Web will raise many compelling, and sometimes problematical, issues. This research paper suggests both a new algorithm that uses session log analysis for predicting end-users' interaction with the Web and a new paradigm for decreasing perceptible latency by applying microprocessor branch prediction algorithms to session logs of a networked community.