finds.dev

guyulongcs/Awesome-Deep-Learning-Papers-for-Search-Recommendation-Advertising

Awesome Deep Learning papers for industrial Search, Recommendation and Advertisement. They focus on Embedding, Matching, Pre-Ranking, Ranking, Post Ranking, Relevance, LLM and RL. Please cite our paper "Deep Learning to Rank in Industrial Search Engines, Recommender Systems, and Online Advertising - An Overview and New Perspectives" (TOIS 2026).

View on GitHub

★ 2,626Pythonupdated Sep 2026

// the take

A reading list of deep learning papers for industrial search, recommendation, and ads ranking systems, organized by pipeline stage (embedding, matching, pre-ranking, ranking, post-ranking, relevance, LLM, RL). Useful for engineers or researchers building large-scale ranking systems who want the landmark papers from Alibaba, Google, Meta, Bytedance, and similar companies without hunting through conference proceedings.

Organized by actual pipeline stage rather than by date or venue, so it maps onto how production ranking systems are architected — you can jump straight to the matching or pre-ranking section instead of filtering a flat list. Heavily weighted toward papers describing systems that ran at real industrial scale (Taobao search, YouTube ranking, Facebook embedding retrieval) rather than academic benchmarks, which matters if you're trying to replicate something that actually works in production. Tracks the recent shift to LLM-based and generative retrieval (TIGER, OneRec, GR) alongside the older CTR-prediction lineage (DIN, DIEN, DCN), so it isn't frozen on 2018-era architectures. A light tagging convention (single/double asterisks) flags the higher-impact papers, which helps when facing 300+ entries.

It's a bibliography, not a toolkit — no code, no reference implementations, no notebooks, just links to PDFs. If you want to actually try one of these architectures you're on your own. The PDFs are hosted directly in the repo rather than linked to arXiv or the publisher, which is a real copyright exposure for ACM/IEEE-owned papers and a DMCA takedown risk that would wipe out large chunks of the list overnight. No annotations beyond title, venue, and year — you still have to open and skim each paper to know if it's relevant to your problem; a one-line note per entry would save a lot of time. As a link list, the usual repo health signals (stars, last push, forks) say nothing about whether the selections are still current or just accumulating — there's no way to tell from the repo itself whether a given subsection has fallen behind the field.

We read the README, the directory tree and a key source file in November 2026. Nobody pays to be listed. How we judge →

// at a glance

Stars
2,626
Forks
298
Language
Python
Last commit
19 Sep 2026

// nearby

If Awesome-Deep-Learning-Papers-for-Search-Recommendation-Advertising isn't quite it.

All repos →