You are on page 1of 2

What is collaborative filtering?

Collaborative Filtering is a technique used when it comes to building intelligent recommender


systems that can learn to give better recommendations as more information about users is
collected. It is an example of how algorithms can leverage data from the crowd. Information
from a lot of people online is collected and used to generate personalized suggestions for any
user. Most websites like Amazon, YouTube, and Netflix use collaborative filtering as a part of
their sophisticated recommendation systems. You can use this technique to build recommenders
that give suggestions to a user based on the likes and dislikes of similar users.

 What are popular collaborative filtering methods? Briefly elaborate on these methods.

Slope One-is nonparametric. Slope One algorithm uses a simple linear regression model to solve
the data sparsity problem. Combined with users’ similarities, the k-nearest-neighborhood method
can optimize the quality of ratings made by users participating in prediction. Based on Slope One
algorithm, a new collaborative filtering algorithm combining uncertain neighbors with Slope One
is presented. Firstly, different numbers of neighbors for each user are dynamically selected
according to the similarities with other users. Secondly, the average deviations between pairs of
relevant items are generated based on ratings from neighbor users. At last, the object ratings are
predicted by the linear regression model. Slope one is a purely predictive nonparametric model
that does not consider any characteristics of the customer or the product when making
recommendations.

Ordinal logistic regression- The second method, which is the parametric model. In case of a
binomial logit model predicts whether a customer will purchase a product or not, and a
multinomial logit regression determines which of several brands a customer might buy, an
ordinal logit model predicts the rating a customer would give to each of the brands under
examination. The principle of the ordinal logit model is to link the cumulative probability of a
level to explanatory variables.  The model considers the attributes of each item in the data set to
understand what fuels consumer ratings, and then it makes predictions based on similarities.

Alternative Least Squares- The alternative least-squares technique is a robust collaborative


filtering method that provides accurate, personalized predictions. Alternating Least Square
(ALS) is also a matrix factorization algorithm and it runs itself in a parallel fashion. ALS is
useful at solving scalability and sparseness of the Rating data, and it is simple and scales well to
very large datasets.

How is collaborative filtering used in Internet companies such as Amazon, Spotify, iTunes?

The “classic”, and still widely used approach to recommender systems based on collaborative
filtering used by Amazon, Netflix, Spotify,iTunes, and YouTube, uses either User-User or Item-
Item relationships to find similar content. The user interaction data is the data we gather from the
weblogs. The item data consists mainly of an item’s features. In YouTube’s case that would be a
video’s meta.  Other data sources could be external data for example, Netflix might add external
item data features such as box office performance or critic reviews. Amazon was probably the
first company to leverage item-to-item collaborative filtering. LinkedIn even developed a
horizontal collaborative filtering infrastructure, known as Browse maps. This platform enables
rapid development, deployment, and computation of collaborative filtering recommendations for
almost any use case on LinkedIn. According to Netflix, the goal of Recommender Systems is to
present several attractive items for a person to choose from. This is usually accomplished by
selecting some items (candidate generation) and sorting them (ranking) in the order of expected
enjoyment (or utility).

You might also like