For companies like Amazon, Netflix, and Spotify, recommender systems drive significant engagement and revenue
Recommender systems provide a scalable way of personalising content for users in scenarios with many items
Recommender systems ... are a subclass of information filtering system that seek to predict the 'rating' or 'preference' that a user would give to an item.
Recommender is a misnomer – discovery assistant is better
System means elements like presentation are important
Recommender systems are systems that help users discover items they may like
Given a matrix of preferences by users for items, predict missing preferences and recommend items with high predictions
Given user preferences for items, recommend similar items based on item content
Recommend items that are liked by friends, friends of friends, and demographically similar people
Recommend items that match the user's current context
They care, but not that much...
Our business objective is to maximize member satisfaction and month-to-month subscription retention... Now it is clear that the Netflix Prize objective, accurate prediction of a movie's rating, is just one of the many components of an effective recommendation system that optimizes our members' enjoyment.
|Transparency||Explain how the system works|
|Scrutability||Allow users to tell the system it is wrong|
|Trust||Increase user confidence in the system|
|Effectiveness||Help users make good decisions|
|Persuasiveness||Convince users to try or buy|
|Efficiency||Help users make decisions faster|
|Satisfaction||Increase ease of usability or enjoyment|
Recommend gifts for Facebook friends using liked pages
Optimising each list's ranking and list ordering, while considering device-specific UI constraints, relevance, engagement, diversity, business constraints, and more...
Generated a single list by statically mixing the outputs of the following algorithms:
A static mix of different approaches can get you very far, but there's a better way
Web search is a recommender system for pages that gives high weight to the user's intent/query
When personalising web search, it seems sensible to use collaborative filtering techniques
My Yandex competition experience: matrix factorisation was a waste of time compared to domain-specific methods
Just like data science: