WP-CamemBERT for next location Prediction

CamemBERT is a Natural Language Processing (NLP) model. The current paper introduced a combination of Bertwordpiece tokenizer and CamemBERT that we named WP-CamemBERT, to solve the next location prediction problem. Substantially, we sought to investigate the performance of our proposed model WP-CamemBERT compared to famous models in the domain that are: Markov, Long-Short Term Memory (LSTMs), Convolutional Neural network (CNN2D), and Support Vector Machine (SVM) models. In order to build the next location prediction model, we used a dataset of mobile users made available through the CRAWDAD project. The results showed that WP-Camembert gave significant improvement in the prediction accuracy and surpassed the state of the art models that are: Markov, LSTMs, CNN2D, and SVM. Particularly, WP-CamemBERT enhanced the next location prediction accuracy by at least 3.2%.