Development of Scalable Recommendation System via Autoencoder

Abstract

Filtering, prioritising, and effectively distributing crucial information on the Internet, where there are so many possibilities, is required to address the issue of information overload, which has potentially become a problem for many Internet users.This issue is solved by recommender systems, which sort through enormous amounts of dynamically created data to provide customers with customised content and services.A recommendation system based on a student's profile may be useful to deliver important information on the subject of study.The method of system development that is most well-known is collaborative filtering.The technique utilised to create the most well-known system is collaborative filtering. The sparsity of the training dataset has a few issues, though, that must be resolved.The training dataset's dimension can be decreased using deep learning and the autoencoder technique. Auto encoders are designed to provide neural networks the flexibility to choose the best encoding and decoding techniques for a particular input. An autoencoder can be used to encode any situation where it is useful.