2021 Review of Alibaba Mama Technology: Advertising Algorithms, AI Research, Open‑Source Projects and Conference Papers

On May 13, 2021, the "Alibaba Mama Technology" channel officially launched. In the preceding 237 days the team published 50 original articles covering advertising algorithm practice, algorithm engineering & system construction, intelligent creativity, risk control, data science and other technical fields. Two algorithm projects were open‑sourced and several papers were presented at well‑known AI conferences.

The post then lists a collection of links grouped by topic:

Advertising Algorithm Practice : articles on CTR model slimming, click‑through‑rate modeling, display ad matching architecture, commercial traffic bidding algorithms, etc.

Algorithm Engineering, Engine & System Construction : experiences on deep‑learning‑driven recall optimization, heterogeneous hardware acceleration, large‑scale asynchronous interaction governance, attribution and growth analysis engines, etc.

Intelligent Creativity : dynamic description ads, creative selection, high‑concurrency image rendering.

Risk Control : anti‑fraud algorithms, advertiser arbitrage detection, brand risk management.

Open‑Source : EFLS (federated learning solution) and CurvLearn (curvature learning framework).

Academic Papers : multiple SIGIR 2021 papers (PCF‑GNN, FSCD‑PreRank), KDD 2021 papers (deep‑confidence‑aware ad exploration, neural auction, multi‑agent bidding games, etc.), CIKM 2021 papers (binary‑code hash embedding, AMTL, STAR, AutoHERI, graph‑based ad retrieval, heterogeneous graph learning), and a CVPR 2021 article on efficient GAN training.

Additional sections include the "Alibaba Mama Data Science Series" (online experiments, AB testing, metric design), the "Marketing Science Series" (C.M.O analysis, consumer asset/behavior analysis, audience communication, channel attribution), and "Engineer Growth" stories for new graduates.

The post ends with a thank‑you message for 2021 and wishes for an interesting 2022.