Let's talk about Recommendation Systems!
summary
Covers how recommendation systems go from a sparse user‑item matrix to filled predictions using matrix factorization (SVD/PCA) after the Netflix Prize jumpstarted research and produced ~10% improvement over Netflix’s baseline, contrasts content‑based and collaborative filtering, explains cold‑start fixes (ask users, metadata), describes hybrid and group/reciprocal recommenders, notes modern stacks combine neural feature extraction with collaborative models (TikTok/YouTube), recommends the Surprise Python library for hands‑on experiments and emphasizes implicit feedback and editorial injection to manage surprise.
AI-written summary · about
chapters (7)
- 0:00Start
- 4:04Recommendation Engines are taking over the world!
- 11:49Human-generated data is temporal by nature
- 12:57Collaboration Filtering
- 14:24Matrix Factorization
- 18:42Hybrid Systems are most popular ones
- 19:41Reciprocal Recommender Systems
related talks
37mA primer on recommendation systems.
20mBuilding Recommender system
01h 31mHands on - Build a Recommender system
01h 38mBuilding Recommender Systems Using Python
33mPicking What to Watch Next: Build a Recommendation System
58mRecommender Systems – A Walkthrough
29mFurther enhancements of recommender systems using deep learning
41mRecommender Systems: Crafting Effective Personalised Experiences