SEARCH · ADS · RECOMMENDATION

Understand the system.
Test every assumption.

Learn search, advertising and recommendation through original sources, short experiments, and evidence you can inspect. One shared foundation. Three connected fields.

Read here, in your browser. Clone the Lab when you are ready to run experiments.

63notebook lessons
4learning paths
EN / 中文two reading languages
56original-material entries

CHOOSE A PATH

Learn in the order the ideas connect.

New to the fundamentals? Begin with chapters 31–34.

FOLLOW THE EVIDENCE

Know what each result can tell you.

The curricula connect author lectures, textbooks and research to independent teaching notebooks. Every path explains its sources and boundaries. A course reference is not a claim of endorsement or a full reproduction.

How the foundations are sourced →

Constructed mechanisms

Small inputs make shapes, gradients, losses and decisions visible.

Real-data experiments

Public logs and benchmarks, with explicit splits, metrics and limitations.

Independent simulations

Explore auction and delivery behavior without calling simulated revenue real revenue.

READ ONLINE. EXPERIMENT LOCALLY.

Ready to change a variable?

This site shows the teaching text, code and saved outputs. Clone the same Lab for editable notebooks, model training and the interactive workbench.

The minimal setup installs the locked environment and notebook kernel. Real-data experiments need their separate domain preparation. The full setup is tested on Apple Silicon macOS; see the guide for platform requirements.

Setup guide and domain requirements
YOUR TERMINAL
git clone https://github.com/lynnlee1229/search-recommendation-ads-lab.git
cd search-recommendation-ads-lab
./setup.sh --minimal
./lab.sh up

Open the local URL printed by the Lab. The website never starts a training job.

Make the next lesson clearer.

Found an unclear explanation, a broken source or a useful experiment? Share a reproducible issue or contribute a focused improvement.

Contribute on GitHub