Future of Movie Recommendations
Media-services providers and production companies such as Netflix, Hulu, HBO Now, etc. have been growing steadily every year since they have been founded. The number of subscriptions on these subscription-based streaming services have been increasing and the companies plan on increasing their earnings in the years going forward. One of the many reasons people fall in love with these services and remain loyal customers is due to their movie recommendations. In this paper, I would like to discuss the strategy that Netflix uses in the recommendation system and how that contributes to their success. Following, I would like to present my theory on the future of movie recommendations.
- Netflix uses a combination of three things to recommend movies to their subscribers.
1. Machine Learning — The first recommendation tactic that Netflix uses is machine learning, a branch of artificial intelligence that studies computer algorithms. Machine learning builds a model based on sample data and makes predictions and decisions based on that “training data”. The main idea and benefit of machine learning is that it improves automatically through experience.
2. Data — The second tool that Netflix uses to recommend movies or tv shows to their users is the use of preexisting data from their users. This is the data that Netflix gathers after their users watch something. Netflix stores the movies you watched and the data collected from other Netflix users who have similar interests to you. It links all the information that they gather about the titles, genre, actors, release year, director and so many more categories. Interestingly, Netflix does not take demographics such as age and gender into account.
3. Algorithms — The last tactic that they use to recommend movies/tv shows is algorithms, a process of set and defined rules which a computer follows. These algorithms are always being re-trained with every visit to the Netflix website to make sure they are as accurate as possible. The data, algorithms and machine learning all work together to complete Netflix’s Recommendation System. They give you brand new suggestions every day to make sure that the movies that they suggest to you are accurate.
- Machine Emotional Intelligence
There are many API software’s online that can read a person’s face and detect their emotions. These software’s and the algorithms they apply, use “facial detection” and “semantic analysis” to decipher a person’s mood whether from a picture, video, text message or an audio file. These API’s use a combination of psychology and technology to combine human emotions into seven different categories. The software will monitor the face “and sense micro expressions by analyzing the relationship between points on the face, based on curated databases compiled in academic environments.” In writing, the software can use “sentiment analysis” to scan the document for keywords that will determine whether the document is positive or negative. Lastly, the software uses “sonic algorithms” to analyze a voice recording which focuses on both the tone and the content of the speech. In conclusion, these algorithms that are linked to the APIs can detect the emotions of a user of an application.
- How will machine emotional intelligence affect how movies are recommended to us?
My theory is that streaming services will no longer just base their recommendation system on machine learning, data and algorithms. Instead they are going to add a fourth category to the mix which is machine emotional intelligence. This feature will use the camera on a computer and monitor the facial reactions of the user as they are watching a movie or TV show. The API will store their emotions in a database and can properly recommend movies based on their previous reactions to shows that they have watched.
As an example, let’s say that a user wants to watch a comedy. Over time, the API software will accumulate a lot of information based on the emotional reactions of the user and can recommend a movie that goes along with his or her sense of humor. It will keep track of the times the user has laughed while watching movies and make suggestions based on that. Additionally, the camera can see how many times and at what points of movies the user gets distracted and picks up their phone. The machine learning algorithm will know not to suggest movies that are likely for the user to get bored and distracted. Additionally, it will not only be able to recommend comedies with the user’s favorite actors/actresses, but also comedies that go along with the user’s sense of humor. The API will know exactly how many times and at which times in the movie the user will laugh.
Furthermore, another way for the emotional intelligence machine to understand how we react to movies and specific scenes in movies is through voice recognition. Often when people watch movies with their families or friends they will comment out loud on the action that is being portrayed on the screen. Or they will use audible reactions such as a gasp or a sniffle depending on their emotions from the movie or TV show. This API software will use “sonic algorithms” to analyze the emotions of the viewer and store this information in a database to help predict which movies a user would like to see next. Using API’s with “facial detection”, “semantic analyses” and “sonic algorithms” streaming services will be able to recommend movies that are tailored directly to the user based on their emotions to movies and tv shows that they have already seen.
Machine learning is a field that has constantly been growing since it was first introduced. Software engineers and data scientists are always looking for ways to predict what will happen and users of applications are listening to the predictions and recommendations of the software to make their next decision. As machine learning gets more accurate, people will rely on it to make even bigger recommendations for them. Streaming services already use machine learning to recommend new movies to their users, but with the combination of emotion recognition APIs, applications will be able to make recommendations with 100% certainty.
Sources:
“Machine Learning textbook”. www.cs.cmu.edu. Retrieved 2020–05–28.
“Deep Learning”. www.deeplearningbook.org.
https://interestingengineering.com/how-exactly-does-netflix-recommend-movies-to-you
