Recommendation Algorithms assist users so that they are able to identify intriguing material within a big corpus. Search engines and other recommendation systems have emerged as efficient methods for locating pertinent information in a reduced amount of time, thanks to the meteoric rise in the quantity of digital resources available on the Internet. Users get extremely helpful assistance from smart recommendation systems and sophisticated search engines these days, which is why it is necessary to develop these tools. The capacity of these systems to retrieve relevant information from vast volumes of data is largely responsible for both their widespread adoption and their practical applicability. As a result, massive and successful corporations have algorithms in place to efficiently learn the user's likes and dislikes and offer the user with the products that they would be interested in purchasing or viewing. Traditional methods such as collaborative filtering techniques and content-based techniques provide impressive results. However, deep learning, which is now being steadily used in domains such as NLP, image and audio processing amongst many other fields, may give answers that are both significant and effective to the issue of information overload. Although there is a vast and extensive body of research on Recommender Systems and potential algorithms to achieve optimal performance, there are very few studies that review generative, discriminative, and hybrid deep learning models and their applications in Recommendation Systems. The aim of this study is to review the various deep learning methods that exist in peer-reviewed literature, analyse their effectiveness, and also look at the growth in Industrial recommender systems over the years.

Recommendation Systems, Deep Learning, Content-Based Techniques, CF, Auto Encoders, Boltzmann Machines, CNN, RNN, Neural Networks, BaRT