Use of Deep Learning in Modern Recommendation System: A Summary of Recent Works

17 Use of Deep Learning in Modern Recommendation System: A Summary of Recent Works Ayush Singhal Contata Solutions LLC, Minneapolis, Minnesota, USA

Pradeep Sinha Contata Solutions LLC, Minneapolis, Minnesota, USA

Rakesh Pant Contata Solutions LLC, Minneapolis, Minnesota, USA

ABSTRACT With the exponential increase in the amount of digital information over the internet, online shops, online music, video and image libraries, search engines and recommendation system have become the most convenient ways to find relevant information within a short time. In the recent times, deep learning’s advances have gained significant attention in the field of speech recognition, image processing and natural language processing. Meanwhile, several recent studies have shown the utility of deep learning in the area of recommendation systems and information retrieval as well. In this short review, we cover the recent advances made in the field of recommendation using various variants of deep learning technology. We organize the review in three parts: Collaborative system, Content based system and Hybrid system. The review also discusses the contribution of deep learning integrated recommendation systems into several application domains. The review concludes by discussion of the impact of deep learning in recommendation system in various domain and whether deep learning has shown any significant improvement over the conventional systems for recommendation. Finally, we also provide future directions of research which are possible based on the current state of use of deep learning in recommendation systems. General Terms Machine Learning, Survey, Recommender Systems Keywords Deep Learning, Recommender system, Literature review, Machine Learning, Collaborative filtering, Hybrid system.

1. INTRODUCTION Our day to day needs ranging from shopping items, books, news articles, songs, movies, research documents and other basic things have flooded several data-ware houses and databases both in volume and variety [1-2]. To this end, intelligent recommendation systems and powerful search engines offer users a very helpful hand. The popularity and usefulness of such systems owes to their capability to manifest convenient information from a practically infinite storehouse[3]. Thus recommendation systems such as Amazon, Netflix and similar others take initiative to know user’s interest and inform users about the items of their interest. Although these systems differ from each other according to the application they are used for, the core mechanism of finding items of user’s interest is that of user’s interest to item matching[4]. In general, recommendations can be generated based on user preferences, item features, user-item transactions, and other environmental factors such as time, season, location. In recommendation literature these are categorized into three primary categories: collaborative filtering (using only the user-item interaction information for recommendation), content based (using user preferences, item preferences or both) and hybrid recommendation models (using both interaction information as well as user and item metadata)[5]. Models under each of these categories have their own limitations such as data sparsity, cold start for users and items[6].
   Given the recent advances in the field of deep learning in various application domains such as computer vision and speech recognition, deep learning has been extended to the area of information retrieval and recommendation systems also[7]. The general opinion about the impact of integrating deep learning into recommendation system is that of significant improvement over the conventional models. In this reviews, we conduct a systematic summarization of various works pertaining to integration of deep learning into recommendation systems to provide substantial basis for reader to understand the impact and directions of future improvement of recommendation systems using deep learning.
2. APPROACH In this section we describe the approach we used to collect, select and summarize research articles for this review. We used the Google scholar search engine to fetch research articles pool to select relevant papers for our review. We used the following keywords to extract articles: “recommender system deep learning”, “collaborative filtering deep learning”, “recurrent neural network recommender systems”. We also set the time filter to “Since 2013” so that we only find articles within the last 5 years. Google Scholar fetched several articles for each query but we performed a manual selection by scanning paper titles to understand if they were actually about deep learning in recommendation systems. The manual selection left us with 33 articles. Each article was then reviewed for the deep

…(Full text truncated)…