What Is Recommender Systems?

The fundamental goal of a recommender system is to reduce information overload and to provide personalized suggestions that can assist the users in the decision-making process.


What music to listen to (for example Spotify)


What movies to watch (for example Netflix)


What consumer products to purchase (for example Amazon, AliExpress)

Social Networks

Suggest groups to join, people to follow, videos to watch, or posts to like

Types of Recommender Systems

Recommender systems are information search and filtering tools that help users to discover relevant items and to make better choices while searching for products or services such as movies, books, vacations, or electronic products. In a general way, recommender systems are algorithms aimed at suggesting relevant items to users.

Similarity Based

User-User and Item-Item approaches

Content Based

Additional Information About Users and/or Items

Location Based

Location Information

Collaborative Filtering

Memory-based Algorithms

Similarity Based Filtering

The user will be recommended items similar to those he/she enjoyed in the past.

The main characteristics of user-user and item-item approaches it that they use only information from the user-item interaction matrix and they assume no model to produce new recommendations

Content Based Filtering

Content based approaches use additional
information about users and/or items.

It is needed to collect and set some specific features to represent items and users. Then the machine learning or deep learning model will be trained by using the combined features.

Location Based Filtered

These services take advantage of the increasing use of smartphones that store and provide the location information of their users.

This could include recommendations for restaurants, museums, or other points of interest or events near the user’s location.

Collaborative Filtering

Recommendations are made based on items consumed by users whose tastes and preferences are similar to that of the referred user.

Memory-based algorithms act on the entire user/item matrix. They assume that users/items can be grouped together by similarity.model-based techniques use a set of user assessments to generate an estimated model, saving the parameters learned during training. Instead of using similarity measurements, these algorithms are characterized by the creation of models.

Matrix Factorization

It is one of the techniques used in collaborative model-based filtering.

According to a certain number of hidden features the user/item matrix is splitted into users and items matrices and the rating score is predicted. Then the items with the highest rates will be recommended to each user.