NETFLIX?
Netflix (specifically the online site) is a video-on-demand streaming website with around 83 million users worldwide.
HOW DOES IT WORK?
The Netflix website displays all of the movies and TV shows in their database currently available to stream. The website database also stores information about each of the titles so that users are able to determine the genre, plot, actors, etc. of each of the thousands of titles available at a given time. Users are also able to browse by category, or use the search bar to display all of the titles related to one actor, director, or to display a specific title they are interested in. One of the most impressive features on the Netflix website is the ever-changing (and improving) recommendation service. Through this feature, the website is able to list all of the titles a user has viewed, and find the different correlations (all the same actress, same genre, same rating) in order to determine other titles the user may enjoy. Some of my own personal recommendation lists are shown below.
HOW DOES THIS RELATE TO COMPUTER SCIENCE?
The recommendation service on the website uses a complex algorithm in order to determine which titles a user may like. Basically, the website's system takes user input, runs the algorithm for determining their preferences, and outputs the thumbnails and descriptions of titles the user may enjoy. Specifically, the algorithm first used by the Netflix company was called CineMax. "According to Netflix, these predictions were accurate within half a star [of user rating] 75 percent of the time, and half of Netflix users who rented CineMatch-recommended movies gave them a five-star rating." In 2006, however, Netflix created a contest for algorithm developers and computer programmers to create an algorithm that was even more accurate at determining the likes of each of the 83 million users. The newly-developed algorithm had a 10 percent improvement over the original, and automatically updates the recommendation system constantly (as the user rates new titles) in order to create the best binge-watching experience.
Photos and information from:
https://chrome.google.com/webstore/detail/netflix/deceagebecbceejblnlcjooeohmmeldh
http://electronics.howstuffworks.com/netflix2.htm
http://www.techradar.com/news/television/how-netflix-works-400-billion-interactions-per-day-ain-t-easy-1306420
https://en.wikipedia.org/wiki/Netflix
Netflix.com
Netflix.com
This is such an interesting article! I have always wondered how Netflix seems to know which tv shows and movies would be right up my alley. It makes me wonder if Pandora, the online radio, uses a similar algorithm to determine music that their listeners would enjoy. And if so, I wonder whose algorithm is more accurate!
ReplyDeleteThis is a cool post and something I can definitely relate to. I also did a post about how websites predict and give recommendations, but it was very interesting to read about Netflix specifically. I wonder if all the moves and television shows are stored as data and how that is implemented in the algorithm you described.
ReplyDelete