APPLICATION OF RESTRICTED BOLTZMANN MACHINE IN A FILM RECOMMENDATION SYSTEM

• Main Author: Muhammad Nofrizal, Revanda
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/83282
• Institution: Institut Teknologi Bandung

Recommendation systems have become a very important way for aiding users in navigating all of the available information and products across the internet in recent years. Among the various techniques that can be used in recommendation systems, Restricted Boltzmann Machines (RBMs) have become one of the popoular approach due to their ability to capture intricate patterns in data. Restricted Boltzmann Machines, a type of generative stochastic neural network, are adept at learning complex probability distributions over input data. In the context of recommendation systems, RBMs excel at capturing user-item interactions in a dataset, enabling them to make accurate predictions and generate personalized recommendations. Unlike traditional recommendation approaches that rely on explicit user-item ratings, RBMs can effectively handle implicit feedback and sparse data, making them suitable for various real-world applications. The use of RBMs for film recommendations have been wildly popular in sites such as letterboxd, a website that lets user review various film, to archive films that user have watched across the years. RBMs can be used to help user on website such as letterboxd, to be exposed to new films that they have not reviewed yet. By using collaborative filtering as a method to implement film recommendations, RBM can be used efficiently to predict a score based on films that users have watched and reviewed.