Analysis of Deep Learning Methods Used in Tourism Recommender Systems

Recommender systems are widely used users’ lives easier by assisting them in narrowing down their product of choice from the various alternatives. Recommender systems (RSs) are information search and filtering technologies that propose items that may be useful to a user. Several companies use them to advertise their products. This work focuses on comparing deep learning algorithms which are used to develop recommendation systems. Four major algorithms are compared - Restricted Boltzmann Machine, Auto-encoder, Convolutional, Attention, and Recurrent Neural Networks. Our results show when a dataset is used which majorly consists of images, CNN (0.85 accuracy) works perfectly. Among RNN (0.81 accuracy), ABRS (0.59 accuracy), and RBM (0.94 accuracy) - RBM performs the best for the given dataset.