How big companies like Netflix and Amazon are able to suggest you relevant products? - That Computer Engineer


Learn | Teach | Explore

Friday, April 27, 2018

How big companies like Netflix and Amazon are able to suggest you relevant products?

The recommendation system is an important application of machine learning. A recommendation system is known to take a large amount of available data and based on user preference or his earlier ratings to suggest further movies for the user.

Do you know big companies like Netflix, Amazon, etc tend to use this recommendation system to show you the adds or products that will be more reasonable for you and also there is a high chance of the user taking interest in that product? This is really beneficial as having an algorithm that can enable the company to show ads or products which are relevant to the user is an optimum method as well as reduces the amount of money and energy resources spent by a company in marketing.

To make you understand this concept better, I will use one of the techniques.

An Easy explanation


Based on users rating of previous movies, we can deduce the taste of an individual related to movies. How this works is that with each movie there will be a feature attached to it as:
The Pursuit of Happiness
Drama: 8/10
Biography: 9/10
Action: 1/10
Romance: 0/10
Thriller: 2/10
Actor: Will Smith

So, on the basis of these values of features, we will be able to calculate some real number value that can help the algorithm determine what kind of movies (based on the feature values). Remember we humans, as well as the machine, tends to understand better if we get some real number value. We won’t be getting deep into technicality. I only gave you an example so that you can understand the concept better.
Consider three movies:

The Social Network                        The Wolf of Wall Street                               Annabelle
 Drama: 5/10                                    Drama:  6/10                                                    Drama: 7/10
 Biography: 9.5/10                          Biography: 6/10                                             Biography: 8/10
 Action: 4/10                                     Action: 3/10                                                    Action: 2/10
 Thriller: 2/10                                   Thriller: 2/10                                                   Thriller: 3/10
 History: 9/10                                   Romance: 5/10                                               Mystery: 9/10

If you carefully observe the ratings of various features for the movie. The algorithm is likely to pick The Social Network as the best option to show as it follows both the criteria of biography and drama in high rating. Following it will be The Wolf of Wall Street. On the other hand, Anabelle is a horror movie and thus, will be the least option to choose among the three options based on the previously watched movie as The Pursuit of Happiness.

Note: The example used is only for understanding and may or may not be precise. 

This was just an insight into what could be the approach used by the algorithm to suggest its user's new movies.

To explore more, you can search for any of the below methods to learn about the approach:

  1. Collaborative filtering
  2. Content-based filtering

No comments:

Post a Comment