The Role of AI in Netflix's Recommendation Engine
Sat, April 17, 2021

The Role of AI in Netflix's Recommendation Engine

Netflix was launched in 1998, originally distributing DVDs by snail mail. Before it was regarded as one of the most popular and successful streaming platforms in the world, it was first known as the pioneer of the DVD-by-mail service / Photo by: Coolcaesar via Wikimedia Commons

 

Netflix was launched in 1998, originally distributing DVDs by snail mail. Before it was regarded as one of the most popular and successful streaming platforms in the world, it was first known as the pioneer of the DVD-by-mail service. The move that they did then that pushed their company to success was relying on the back catalog of studios and production houses. 

In a nutshell, Netflix approached these studios expressing its desire to buy licenses for older movies and TV shows. Netflix wanted people to order titles from them that they couldn’t find at the local video stores. Meanwhile, studios were able to gain sales from their old titles. It was an ideal setup. Netflix grew strongly for the next couple of years. From only 300,000 users in 2000, it increased to 600,000 in 2002 and 4.2 million in 2005. After five years, it introduced a feature that changed everything: streaming. It also partnered with several tech companies to diversify the ways people could have it.

In 2012, Netflix added another feature that drove its further success: original content. While the first series “Lilyhammer” was released during that year, it was “House of Cards” and “Orange Is the New Black” that made significant impacts. There’s no stopping Netflix since then, growing at a rapid rate. Business of Apps, the leading media and information brand for the app industry, reported that user growth in the third quarter of 2019 reached 78.6 million, up from 58.5 million in Q3 in 2018.

Today, users watch more than 140 million hours of content on Netflix every day. It continues to be the number one choice of people who want to watch movies, series, and documentaries. Analysts projected that the streaming platform could reach 200 million subscribers by the end of 2020 and 300 million subscribers by 2028. Behind these huge numbers is artificial intelligence, a critical technology that Netflix uses to stay on the top. 

Netflix Knows What Users Like

Netflix’s success story can’t be explained without mentioning the huge contribution of AI. The streaming platform not only looks at millions of ratings, “plays,” and searches made by its users but also the entire viewing history of billions of hours of content streamed per month. A perfect example of this is the success of “House of Cards.” Netflix spent more than six years to collect enough viewer data to engineer a show that would become a worldwide success. 

The New York Times even claimed that “Netflix is commissioning original content because it knows what people want before they do.” Today, the streaming service is well-known for its recommendation engine that knows what its users want. According to Forbes, an American business magazine published bi-weekly that features original articles on finance, industry, investing, marketing topics, technology, communications, science, politics, and law, focusing on business, investing, technology, entrepreneurship, leadership, and lifestyle, Netflix’s analysis showed that a typical user will lose interest in 60 to 90 seconds when choosing something to watch. 

Thus, it has come up with a recommendation engine to compel its users to start playing content within those 90 seconds. Carlos Gómez Uribe, the vice president of product innovation at Netflix where he leads the team that develops algorithms for movie recommendations and searches, stated that their personalization efforts aim to help users to find something they will love to watch as soon as they open Netflix. “Knowing that we have 60 to 90 seconds to help you find something great, it is our goal to develop the most personalized experience as possible, based on your unique preferences and tastes, so we can surface the titles you will enjoy as fast as possible,” Gomez-Uribe said. 

Netflix’s effective way of predicting accurate content is all because of big data. Big data can translate huge amounts of data that are being processed by optimized algorithms for shaping better business strategies and outcomes. The streaming platform also uses machine learning to analyze and rank content according to a user’s preferences. According to Muvi.com, an enterprise SaaS platform that allows content owners to launch their own video streaming platform, some of the information that Netflix is using include interactions such as your browsing history or your ratings of titles, relevant information like genre, categories, titles, the cast and other users’ interactions with similar tastes. 

Also, Netflix considers other factors to personalize your viewing experiences such as the time of the day you usually watch content, the devices on which you watch them, and the usual time duration of your watching hours. For instance, if you liked watching “Stranger Things,” Netflix will recommend “Black Mirror” for you to watch. Also, Netflix’s recommendation algorithms can recognize what shows or movies you have completed. They also place the row of new titles that a user would likely to enjoy next at the very top of the page.

The streaming platform not only looks at millions of ratings, “plays,” and searches made by its users but also the entire viewing history of billions of hours of content streamed per month / Photo by: Stock Catalog via Flickr

 

Simplifying Data Science and Machine Learning Workflows

In recent years, we have witnessed how tech companies released open-source machine learning tools to train their AI and machine learning algorithms. For instance, Facebook has Pythia, a deep learning framework for image and language models, and Uber has Ludwig, a toolbox built on top of Google’s TensorFlow machine learning framework. Recently, Netflix shared its new and freely available multi-language programming notebook, Polynote. It integrates with Apache Spark and offers robust support for Scala, Python, and SQL.

According to VentureBeat, an American technology website headquartered in San Francisco, California and publishes news, analysis, long-form features, interviews, and videos, Polynote was designed to enable data scientists and AI researchers to integrate Netflix’s JVM-based machine learning framework with Python machine learning and visualization libraries. The streaming platform believes that it has great potential to address the needs of the company. 

“On the Netflix personalization infrastructure team, our job is to accelerate machine learning innovation by building tools that can remove pain points and allow researchers to focus on research. Polynote originated from a frustration with the shortcomings of existing notebook tools, especially with respect to their support of Scala,” the company said. 

AI has a huge potential in making Netflix even more successful. Thus, we can expect that the streaming platform would bring us more unique services in the future.