This article shares a step‑by‑step experience report for newcomers to the Tencent Advertising Algorithm competition, covering how to obtain the dataset, understand the problem, build simple rule‑based baselines, choose models and features, and apply practical tricks to achieve scores above 85.
The week‑1 champion of the 2019 Tencent Advertising Algorithm Competition shares practical experience on data cleaning, feature engineering, model selection (including LightGBM and deep learning), validation strategies, and tips for handling massive ad exposure logs to achieve high SMAPE and monotonicity scores.
This article explains how real‑time bidding works in computational advertising, defines the budget‑constrained bidding problem, models it with reinforcement learning, and presents a deep‑network implementation together with visual analysis and key references.
This article describes the business background, requirements, multi‑layer architecture, implementation details, usage workflow, encountered issues, and future plans of PCMA, a platform that automates the validation of advertising placements across PC, mobile web, and app environments.
This article introduces the Deep Interest Evolution Network (DIEN), an advanced deep learning model that extracts and evolves user interests over time to improve click‑through rate prediction for display advertising, detailing its background, architecture, auxiliary loss, attention‑augmented GRU, and both offline and online performance gains.
The 58 Group technical salon presented detailed engineering practices on video recommendation system architecture, image creative optimization for search ads, a programmatic creative platform, and a Kubernetes‑based deep learning online prediction service, highlighting micro‑service design, distributed indexing, and real‑time model deployment.
Alibaba's X-Deep Learning (XDL) framework, the first open‑source deep learning system designed for high‑dimensional sparse data, powers advertising, recommendation, and search workloads, delivering industrial‑scale performance and boosting revenue while offering an open, easy‑to‑use solution for the broader AI community.
iQIYI uses two Lookalike audience extension methods—tag‑based using weighted tag scoring and supervised machine‑learning using logistic regression with engineered DMP and ad behavior features—both improving ad performance, e.g., 20% higher Trueview completion and up to 60% lower conversion cost.
The IJCAI 2018 International Advertising Algorithm Competition was won by JD.com algorithm engineer Hua Zhixiang, who employed a two‑stage LightGBM model with transfer learning and carefully designed statistical, temporal, ranking, and representation features to achieve top conversion‑rate predictions on massive e‑commerce advertising data.
The team from Jilin University, Harbin Institute of Technology, and Beijing University of Posts and Telecommunications shares their winning strategy for the Tencent Advertising Algorithm Competition, detailing their feature engineering, model selection, and future work to handle large‑scale data challenges.
The article examines how blockchain technology enables decentralized data marketplaces that let individuals reclaim control over their personal information, monetize it through smart contracts, and challenge the data‑driven advertising dominance of major tech giants such as Google and Facebook.
This article recounts Alibaba‑Mama researcher Jing Shi’s presentation on the evolution of deep learning for click‑through‑rate (CTR) estimation, covering the shift from handcrafted features and linear models to piecewise linear MLR, end‑to‑end neural networks, multi‑interest user modeling, and large‑scale distributed training challenges.
The article shares the runner‑up team’s reflections on the first Tencent Social Advertising university algorithm contest, covering data splitting, feature engineering, handling large datasets, model selection, ensemble techniques, and final advice to help future participants succeed in conversion‑rate prediction challenges.
The article introduces a commercial testing platform used by an advertising team, detailing its architecture, core components, monitoring and scheduling mechanisms, current advantages and shortcomings, and outlines planned enhancements to improve data construction, result recording, and anti‑fraud coverage.
The article shares a Beijing University team's experience in the first Tencent Social Advertising algorithm contest, detailing their fourth‑place finish, best‑presentation award, and five key strategies—including business‑logic analysis, model innovation, multi‑model fusion, teamwork, and leveraging existing research—to improve conversion‑rate prediction performance.
This article explains the architecture and key technologies of 360’s SSP advertising engine—including flexible micro‑service layers, DAG‑based topology, rule and template management, and high‑performance Go‑based HTTP frameworks—that enable billion‑scale ad delivery with low latency and high concurrency.
This article explores how the wildly successful game PUBG illustrates various internet business models, explains the components of a business model, examines common ecosystem and freemium models, details advertising and revenue streams of 58.com, and shares design strategies for large‑scale B‑end products.
Meituan‑Dianping’s DSP coarse‑ranking filters large ad candidate sets by scoring ads with user‑profile, weather, and keyword scenario models—using frequent‑itemset mining, AdaBoost, and TF/IDF—then aggregates these scores via a linear‑regression model to select high‑relevance ads for fine‑ranking, boosting click‑through and conversion rates.
Senior Alibaba algorithm expert Jing Shi shares his six‑year journey from joining the company, the challenges of large‑scale ad click‑through‑rate modeling, the evolution from linear logistic regression to deep learning, and practical advice for aspiring AI engineers and interview candidates.
This article explains the technical background of click‑through‑rate (CTR) prediction, critiques traditional linear models, introduces Alibaba’s Mixed Logistic Regression (MLR) algorithm with its advanced features and large‑scale distributed implementation, and reviews its successful deployment and remaining challenges in advertising systems.
The article introduces the Tencent Social Ads university algorithm contest weekly champion team Groot, details their background, and outlines their practical machine‑learning approach—including training set construction, XGBoost model selection, and feature engineering—while encouraging broader participation in such competitions.
This article introduces two Python baselines—one using historical average conversion rates and another applying one‑hot encoding with logistic regression—to demonstrate feature engineering and model training for an app advertising conversion rate prediction contest.
The expert live Q&A session for Tencent's Social Ads algorithm competition was recorded and the full FAQ, covering data labeling, feature handling, model selection, imbalance mitigation, and practical engineering tips, is presented here for participants and researchers alike.
Meituan’s demand‑side platform combines real‑time bidding with a two‑service architecture—RecServer for multi‑scenario ad recall and PredictorServer for CTR and conversion‑value prediction—leveraging behavior, location, collaborative‑filtering and matrix‑factorization features, logistic‑regression and GBDT models, and continuous A/B and metric monitoring to optimize ROI.
This article examines the architecture of Weibo's high‑availability advertising platform, covering match service design with OpenResty, index sharding, business logic optimization, dynamic auto‑scaling, and a real‑time monitoring pipeline to ensure stable, high‑performance ad delivery at massive scale.
This article presents the research and implementation of a high‑performance, highly available OTT advertising platform, detailing its system architecture, key technologies such as multi‑domain routing, containerized modules, two‑level caching, double‑layer Boolean retrieval, dynamic CDN scheduling, graceful degradation, and comprehensive disaster‑recovery strategies.
This article explains how Hulu’s advertising platform leverages big‑data pipelines, time‑series forecasting, bipartite‑graph optimization, linear‑model targeting, and real‑time allocation to improve sales, delivery, reporting, and planning of guaranteed‑delivery brand ads.
Last month Ctrip Technology Center hosted a deep‑learning meetup featuring academic and industry experts from UCL, Fudan, Southeast University, Nanjing University, Huawei, Sogou and others, who presented real‑world applications of deep learning in travel, advertising, natural language processing, computer vision, and knowledge graphs.
This article presents a deep‑learning‑based approach to multi‑field categorical data, explains FM and FNN embeddings, compares several click‑through‑rate prediction models on Criteo and iPinYou datasets, and demonstrates that factorisation‑machine‑supported neural networks significantly outperform logistic regression and other baselines.
This article explains the principles of Factorization Machines (FM) and Field‑aware Factorization Machines (FFM), their implementation details, and how Meituan‑Dianping applied FFM in a DSP platform to achieve superior CTR and CVR estimation for sparse, high‑dimensional advertising data.
The article compiles detailed notes from the 2015 Internet+ Big Data Summit, highlighting how data monetization reshapes advertising, drives financial analytics, improves operational efficiency, and strengthens security, while presenting real‑world case studies, models, and practical recommendations from industry experts.
This article records a WeChat Q&A where industry experts discuss selecting open‑source big data solutions, advertising‑specific data scenarios, and share a practical lambda‑style platform architecture featuring Hadoop, Spark, Storm, Elasticsearch, Redis and MySQL.