Pathway Prediction Using Similar Users and the N-gram Model

This paper is about our research on user pathway prediction for being applied to a location aware system. In particular, we propose a prediction method based on an N-gram model with Kneser-Ney smoothing (KNS), originally developed by other researchers for statistical language model smoothing, and introduce the use of the transition information of similar users into KNS. We then verify the performance of the proposed prediction method by comparing it with an existing prediction method and a prediction method based on KNS using all users' information. The comparison result reveals that the proposed method outperforms its counterparts on all performance metrics: precision, recall, F-measure, and CA.