A Chinese-language survival guide for ad-tech ML

What it does This repository is a curated knowledge base that collects learning materials for recommender systems, computational advertising, and CTR/CVR prediction. It organizes Chinese-language tutorials, landmark papers, open-source tools, and competition solutions into a single roadmap-style index. The maintainer filters for engineering-oriented research from Google, Alibaba, Facebook, Airbnb, and other production-scale environments.

The interesting bit Instead of dumping links, the collection follows a practical learning arc from statistical models—logistic regression, GBDT, FFM—through deep-learning architectures such as Wide & Deep and DeepFM, and into real-world competition post-mortems. Several classic papers are mirrored directly as PDFs, so the repo doubles as a mini archive.

Key highlights

• Direct PDF mirrors of industrial classics including Google’s Wide & Deep, Facebook’s GBDT+LR, Huawei’s DeepFM, Airbnb’s embedding-based search ranking, and Alibaba’s ESMM multi-task model
• Tooling index covering LightGBM, XGBoost, LIBFFM, xLearn, and DeepCTR
• Annotated competition solutions from Criteo, Avazu, IJCAI 2018, and the 2018 Tencent Ads Algorithm Contest
• Companion project RecNews for tracking frontier technical articles
• Claims weekly updates

Caveats

• This is a curated reading list and document mirror, not a runnable framework or library
• The README includes a Travis CI badge that points to kamyu104/LeetCode-Solutions, which appears to be leftover scaffolding
• Nearly all commentary and linked tutorials are in Chinese

Verdict A solid bookmark for Chinese-speaking engineers who want a structured on-ramp to industrial recsys and ad-ranking. English-only readers or those seeking a drop-in code package should look elsewhere.