From Serendipity to Strategy: How Recommendation Systems Revolutionize User Experience
After completing an intriguing novel, a riveting series or a captivating movie, the most common question that arises is what’s next? which movie or show to watch next or which novel to read next? Such questions are easily answerable if one’s field of interest is known.
A recommendation system works on similar rules. A recommender system, also known as a recommendation system, is a type of information filtering system that provides suggestions for items that are most relevant to a specific user. It takes into account the types of shows watched/ books read by the user to provide multiple answers to the what’s next dilemma. Thus, in simpler terms, a recommendation system helps users find compelling content in a large corpus.
Recommendation systems are further classified into 3 categories:
- Collaborative filtering systems
- Content based filtering systems
- Hybrid systems
Collaborative Filtering
Collaborative filtering uses similarities between users and items simultaneously to provide recommendations. This allows for serendipitous recommendations; that is, collaborative filtering models can recommend an item to user based on the interests of a similar user.
Example: If a user has shown interest in a movie A, B and C, then the system will recommend another movie, say X, on the basis of ratings provided by another user who also liked the aforementioned 3 movies (i.e., a similar user).
A real-life application of this scenario can be seen when surfing through the amazon site. When we shop on Amazon it recommends new products saying “Customer who brought this also brought”
Collaborative Filtering is further divided into 2 subclasses:
a. User Based Collaborative Filtering
It finds similar users and recommends products based on what other people with similar consumption patterns liked. It focuses on the Nearest Neighbour approach for recommendations, which looks at the users' rating patterns and finds the nearest neighbors, i.e. users with ratings similar to yours. The algorithm then proceeds to give you recommendations based on the ratings of these neighbors.
b. Item Based Collaborative Filtering
Item based collaborative filtering finds similarity patterns between items and recommends them to users based on the computed information. Item based filtering looks at the similarity between different items, and does this by taking note of how many users that bought item X also bought item Y. If this correlation is high enough, a similarity can be presumed to exist between the two items, and they can be assumed to be similar to one another. Item Y will from there on be recommended to users who bought item X and vice versa.
Content Based Filtering
This type of recommendation system displays relevant items based on the content of previously searched items by the users. The attribute of the product that the user likes is referred to as content in this context. Products are tagged with specific keywords in this type of system.
Example: If a person likes a movie, say A, belonging to thriller genre, sub-genre — crime, then the system will recommend the user movies with tag of ‘crime’ and ‘thriller’
A real-life example of this can be seen on the Netflix platform. After watching a show/movie, it presents the user with a section as ‘More like these’ which showcases series/movies of similar genre.
Hybrid Systems
A hybrid recommendation system is a type of recommendation system that is a combination of content and collaborative filtering methods. Combining collaborative and content-based filtering may help overcome the limitations we face when using them separately, and it may also be more effective in some cases. Hybrid recommender system approaches can be implemented in a variety of ways, such as using content and collaborative-based methods to generate predictions separately and then combining the predictions, or simply adding collaborative-based method capabilities to a content-based approach (and vice versa).
There are two types of hybrid recommendation systems: parallel and sequential. The parallel design feeds multiple recommendation systems, and each of those recommendations is combined to produce a single output. The sequential design sends the input parameters to a single recommendation engine, whose output is then passed on to the next recommender in the chain.
Hybrid systems combine different models to combat the disadvantages of one model with another. This overall reduces the weaknesses of using individual models and aids in generating more robust recommendations. This yields more robust and personalized recommendations for users.
A real-life example of this technique is Netflix. It employs a hybrid recommendation system in which they generate recommendations to users based on similar users’ watch and search habits (collaborative filtering) in conjunction with movies that share similar characteristics and have been rated by users (content based).
The code for recommender systems can be found here.
Thanks for reading! If you have any queries, feel free to reach me on Gmail or my LinkedIn Profile or GitHub profile.
