Modified node2vec and attention based fusion framework for next POI recommendation

The rise of location-based services has led to the widespread adoption of location-based social networks (LBSNs), which play a vital role in making recommendations for the next Point-of-Interest (POI). This paper introduces a modified node2Vec and attention-based fusion framework for the next POI recommendation. We start by preprocessing the raw data to gather the relevant information and present a modified node2vec algorithm to generate the feature vectors for users and locations. These feature vectors are then processed using the attention-based framework. The processed features are then used to create well-labeled and balanced datasets which are grouped by specific time intervals. These datasets are then used for training various ML classifiers which are ensembled in a weighted manner to make an improved fusion based recommendation system. The intensive experimental simulations demonstrate the effectiveness of the proposed framework over existing state-of-art methods.