A Study on Autoencoder based Technique in Modern Movie Recommendation System

In today's era, people are suffering from some common problems like stress, depression, anxiety, discouragement and pessimism due to busy work life. Thus, a sprinkle of entertainment in life is a basic necessity to boost lives with some fresh energy. From long ago movies are considered as a great source of entertainment. The immense growth of the internet, mobile devices, and e-business lead to the tremendous growth of data. As a result, the searching process becomes very tedious and time consuming, which leads to the development of a system that can filter out information based on the user's requirement and priority. Recommendation systems are an effective information filtering tool, which constantly observes the behavior of users and provides recommendations according to their interests and preferences. This paper presents a preliminary survey on various movie recommendation approaches like content-based filtering, collaborative filtering, and hybrid approach. It also discusses various drawbacks of these approaches and mainly focuses on an autoencoder based deep learning approach to enhance the performance of recommendation systems.