Chapter 2 · Machine learning basics
Recommendation systems
Recommenders guess what you will like from what you and people like you chose before.
Chapter 2 · Machine learning basics
Recommenders guess what you will like from what you and people like you chose before.
People rated some items; many cells are blank…
Nobody has rated everything. The recommender learns who is like you, then guesses your missing ratings from theirs.
Imagine a table: people down the side, items across the top, and a rating where someone has tried something. Most cells are empty because nobody tries everything.
Tap a dish to cycle its rating from 1 to 5, then back to not rated. Rate at least two dishes that others have rated.
Simplified teaching version: five made-up people, eight dishes, and one similarity measure.
How does a collaborative recommender guess a rating you have not given?
A brand-new user has rated nothing. What is the problem?
What is a risk of showing only what similar people liked?