Getting Started With Recommendation System

Photo by Chris Ried on Unsplash

What is Recommender System

Recommender systems, in a broad sense, are algorithms that aim to suggest relevant items to users (items being movies to watch, text to read, products to buy, or anything else depending on industries).

In this article, we’ll look at different recommender system paradigms. We’ll explain how they work, describe their theoretical foundations, and discuss their strengths and weaknesses.

Paradigms or Filtering Strategies

The two major paradigms of recommender system:

• Collaborative
• Content-based

There is also a method known as the hybrid method.

Content-based Filtering:

This filtration strategy is based on the information about the items that have been provided. The algorithm suggests products that are similar to those that the user has previously liked. This similarity (generally cosine similarity) is calculated based on the information we have about the items and the user’s previous preferences.

Assume we want to create a movie recommendation engine that is content-based then this type of recommendation system uses a user’s current favorite movie as input. The program then examines the movie’s contents (storyline, genre, cast, director, and so on) to identify other films with similar content. The system then ranks similar films based on their similarity scores and suggests the most relevant films to the user.

Disadvantages:

• Different products do not get much attention from the user.
• The model can only be as good as the hand-engineered features.
• The model can only make suggestions based on the user’s current interests. To put it another way, the model’s ability to expand on the users’ existing interests is limited.

Content-based recommendation system

Collaborative Filtering:

Collaborative approaches for recommender systems are ways that provide new recommendations exclusively based on past interactions recorded between users and products.

This algorithm tries to find users who are similar based on their activities and preferences first (for example, both the users watch the same type of movies or movies directed by the same director). Now, if one of these users (let’s call them A and B) has seen a movie that user B hasn’t seen yet, that movie is recommended to user B, and vice versa.

For example, When you shop on Amazon, you’ll often see lists like “Customers who viewed this item also viewed” and if you bought it, “Customers who bought it also bought.”

Advantages:

Collaborative techniques have the advantage of requiring no knowledge on people or goods, allowing them to be employed in a variety of circumstances. Furthermore, the more people engage with goods, the more accurate new recommendations become: fresh interactions logged over time offer new information and make the system more and more effective for a specific set of users and items.

Disadvantages:

• Cannot handle fresh items
  If an item is not seen during training, the system can’t create an embedding for it and can’t query the model with this item. This issue is often called the cold-start problem
• Hard to include side features for query/item

Content-based vs Collaborative Methods

simple representation of Content-based & Collaborative methods

Thanks for Reading!!

References:

Introduction | Recommendation Systems | Google Developers

Introduction to recommender systems