PlatformBid, the first benchmark designed from a unified advertising‑platform perspective, expands real‑time bidding research beyond DSP‑centric single‑advertiser goals by adding platform‑level constraints, three realistic evaluation settings, and a new Flow‑Matching‑based algorithm (BidFlow) that achieves state‑of‑the‑art performance both offline and in a live Kuaishou e‑commerce deployment, delivering a 0.68% consumption lift.
The Kuaishou technology team had 25 papers accepted at the prestigious KDD 2026 conference—including three oral presentations—covering generative recommendation, automated bidding, semantic ID learning, multi‑behavior modeling, and other AI‑driven advances that are already deployed at massive scale on the platform.
The paper presents PlatformBid, the first unified‑platform auto‑bidding benchmark that evaluates both platform‑level revenue and advertiser fairness across three realistic competition settings, and introduces BidFlow, a flow‑matching based bidding algorithm that achieves state‑of‑the‑art performance both offline and in live production.
A comprehensive dialogue between Hong Kong Chinese University’s vice president and Tencent’s advertising AI experts explores AI frontiers, trustworthy and embodied AI, the four‑generation evolution of recommendation systems, the role of the Tencent Advertising Algorithm Competition in talent development, and practical advice for young professionals navigating the AI‑driven advertising industry.
The article traces the decline of interruptive ad formats—from banner blindness to AI‑driven search summaries—showing how AI eliminates the need for disruptive ads and instead enables ultra‑precise, intent‑based matching that serves only the 5% of users who truly want the product.
OpenAI has moved from a limited ad test in February to a full‑scale ChatGPT Ads platform after the GPT‑5.5 Instant release, detailing how it infers user interest through memory, separates ad slots from answers, sets premium pricing, and balances free‑user controls with privacy safeguards.
GR4AD, Kuaishou’s generative recommendation system, redesigns the entire ad pipeline—from tokenizing multimodal ad material to value‑aware learning, lazy decoding, and dynamic beam search—delivering over 4 % revenue lift, higher eCPM, and sub‑100 ms latency for more than 400 million users.
This article dissects the full advertising funnel, explains why CTR and eCPM are critical, and demonstrates how StarRocks combined with Flink can deliver minute‑level real‑time monitoring, material selection, anomaly alerts, A/B testing, and a successful migration from Druid for massive ad‑tech workloads.
The article reviews the 2025 evolution of Tencent advertising’s bidding system—from the second‑generation GMPC control algorithm through the third‑generation MRB reinforcement‑learning model to the fourth‑generation generative RL GRB—detailing architectural upgrades, multi‑channel modeling, training pipelines, and experimental gains, and outlines the 2026 AI‑agent roadmap.
Tencent's ad tech team redesigned its ad ranking system by adding a parallel user‑experience‑optimized pipeline and evolving from manual CEM tuning to DDPG‑based reinforcement learning, achieving 10‑20% improvements in CTR, repeat‑view rates, and other experience metrics while maintaining overall spend.
Bid2X introduces a bidding‑environment foundation model that unifies heterogeneous ad‑bidding data, leverages variable and time attention mechanisms, handles zero‑inflated distributions, and demonstrates superior offline performance across eight large‑scale datasets and significant online gains in GMV and ROI when deployed on a major e‑commerce platform.
RAVEN is a reinforcement‑reasoning framework that combines curriculum learning with hierarchical rewards to enable multimodal large language models to accurately locate and classify violation segments in advertisement videos, even under noisy, large‑scale industrial data.
A recent analysis of the ChatGPT Android test build reveals embedded advertising features, outlines three ad formats, examines OpenAI's mounting financial pressures, and discusses the potential impact on user trust and the broader AI industry.
This article reviews Kuaishou's two‑year exploration of large‑model techniques in advertising, detailing the challenges of content‑domain ad estimation, the use of multimodal and LLM technologies to harness full‑scope user behavior and external knowledge, and the COPE and LEARN frameworks that delivered measurable business gains.
Advertisingknowledge transferlarge language models
This article reviews Kuaishou's two‑year exploration of multimodal large‑model techniques for advertising, outlining challenges in content‑domain ad estimation, the COPE unified product representation framework, and the LEARN LLM knowledge‑transfer approach that together improve ad system performance.
This article reviews Kuaishou's two‑year exploration of large‑model techniques in advertising, outlining challenges in content‑domain ad estimation, introducing the COPE unified content representation framework and the LEARN LLM knowledge‑transfer approach, and showing how these innovations delivered tangible business gains.
This article reviews Kuaishou's two‑year exploration of multimodal large‑model techniques for advertising, detailing challenges of fragmented user behavior, the COPE unified product representation framework, and the LEARN LLM knowledge‑transfer approach that together delivered measurable business gains.
This article reviews Kuaishou’s two‑year exploration of large‑model techniques in advertising, detailing the content‑domain estimation challenges, how multimodal and LLM approaches improve full‑domain behavior utilization and external knowledge integration, and introducing the COPE product‑content representation framework and the LEARN LLM knowledge‑transfer system.
Private Set Operations (PSO) enable multiple parties to perform set intersections, unions, and related computations on encrypted data, preserving privacy through cryptographic techniques such as public‑key encryption, oblivious transfer, and garbled circuits, and are applied across advertising, finance, healthcare, and government for secure data collaboration.
This article introduces LCRON, a novel end-to-end training framework for cascade ranking systems that aligns training objectives with overall recall, addresses stage interaction challenges, and demonstrates significant performance gains on public benchmarks and in Kuaishou’s commercial advertising platform.
JD's retail technology team presents five SIGIR 2025 papers that introduce advanced graph neural, causal optimal transport, domain‑oriented relevance, multi‑objective bid‑word generation, and hierarchical user behavior models to dramatically improve precise matching in e‑commerce search, recommendation, and advertising.
This article showcases three JD tech case studies—using large language models for e‑commerce query expansion, applying sparse large models with scaling‑law experiments to improve ad prediction, and building proactive risk‑prevention systems—to illustrate practical AI engineering that drives higher recall, conversion, and system robustness.
The article describes how a custom algorithm review platform, built around goldenset test cases, quantifies and visualizes CAD recognition and advertising design tool outputs, enabling rapid regression testing, objective metric tracking, and efficient manual review, ultimately improving development speed and bug detection rates.
The article recounts Xiaoting’s journey from a PhD research background to leading JD.com’s ad‑fraud detection, detailing how large language models, reinforcement learning, and model distillation were applied to identify hidden address codes, reduce false‑positive rates to 0.3%, and balance accuracy with real‑time performance in a high‑traffic e‑commerce environment.
The article presents the Universal Recommendation Model (URM), a large‑language‑model‑based recall framework that integrates world knowledge and e‑commerce expertise through knowledge injection and prompt‑driven alignment, achieving significant offline recall gains and a 3.1% increase in ad consumption while meeting high‑QPS, low‑latency production constraints.
JD’s advertising platform replaces rule‑based recall with a generative large‑model pipeline that unifies e‑commerce knowledge, multimodal user intent, and semantic IDs across recall, coarse‑ranking, fine‑ranking and creative optimization, while meeting sub‑100 ms latency and sub‑¥1‑per‑million‑token cost through quantization, parallelism, caching, and joint generative‑discriminative inference, delivering double‑digit performance gains and paving the way for domain‑specific foundation models.
The article presents a series of technical breakthroughs by JD's advertising team that improve the quality and coverage of AI‑generated ad images through a trustworthy multimodal feedback network, introduce a large human‑annotated image dataset, and enhance creative ranking with offline multimodal representations and online architecture optimizations, ultimately achieving more precise and scalable ad personalization.
The paper introduces One4All, a scalable multi‑task generative recommendation framework for CPS advertising that combines few‑shot intent prompting, a Rewards‑in‑Context multi‑objective optimization, and an online model‑selection strategy, delivering 2‑3× offline HitRate/NDCG gains and notable online CTR, CVR, and commission improvements.
In 2024 the advertising team deployed an end‑to‑end AIGC pipeline that automatically creates high‑quality ad images, uses the multimodal Reliable Feedback Network and the million‑size RF1M dataset to filter outputs, builds rich offline and online multimodal representations with contrastive and list‑wise learning, and optimizes ranking architecture to deliver scalable, personalized creative selection.
The 2024 advertising team introduced a suite of AI‑driven techniques—including a trustworthy feedback network, a large‑scale human‑annotated dataset, multimodal large language model representations, and online ranking architecture upgrades—to dramatically improve the quality, coverage, and personalization of generated ad creatives.
MaRCA, a multi‑agent reinforcement‑learning framework, allocates compute across JD’s advertising playback chain by jointly estimating user value, resource consumption, and action outcomes while dynamically adjusting to real‑time load, achieving roughly 15 % higher ad revenue without extra compute resources.
Alibaba Mama’s advertising platform migrated from a monolithic Flink‑Kafka pipeline to a layered Paimon lakehouse, adding DWS upsert support and multi‑layer storage, which delivers minute‑level data freshness, cuts latency by 2.5 hours, reduces resource use over 40 %, halves development effort and achieves ≥99.9 % availability.
The paper presents a generative recommendation framework for JD Alliance advertising that combines semantic‑ID modeling, large‑model pre‑training and fine‑tuning, and Direct Preference Optimization (including Softmax‑DPO and β‑DPO) to jointly boost click‑through and conversion rates, achieving +0.6% UCTR and +8% UCVR in online tests while outlining future multi‑objective extensions.
This article details JD Advertising's practical use of large‑model agents—including RAG, function calls, and workflow automation—to improve service efficiency, monitoring, and developer productivity across its advertising platform.
The NeurIPS 2024 Auto‑Bidding competition attracted over 15,000 submissions and 1,500 teams, featuring two tracks—General and AI‑Generated Bidding—where Kuaishou’s commercial algorithm team secured first place in both by leveraging reinforcement‑learning‑based online exploration and a decision‑transformer‑driven generative approach, achieving more than a 5% lift in ad revenue.
This article presents a comprehensive solution for heterogeneous long‑behavior sequence modeling in advertising recommendation, introducing the TIN backbone, Disentangled Side‑Info TIN, Stacked TIN, and Target‑aware SASRec, along with platform‑level optimizations that enable million‑scale sequences while delivering significant online performance gains.
The JD Advertising R&D team applies cutting‑edge AI techniques—including query intent models, multimodal representation pipelines, reinforcement‑learning‑based auction mechanisms, generative recommendation with quantized product tokens, and large‑model infrastructure—to boost traffic valuation, ad relevance, revenue, and creative generation across the platform.
This article analyzes how large‑model AIGC reshapes the production, evaluation, and deployment of advertising creatives on Xiaohongshu, detailing the business motivations, technical pipeline, controllable generation, reward‑model filtering, and experimental results that balance commercial efficiency with community tone.
This study investigates how adding an auxiliary ranking loss to click‑through‑rate (CTR) models not only improves ranking but also alleviates gradient‑vanishing for negative samples, thereby boosting the primary classification performance, especially under sparse positive‑feedback conditions.
ADSNet introduces an adaptive Siamese network for cross‑domain lifetime value (LTV) prediction in advertising, leveraging external channel data to mitigate sample sparsity, employing gain‑based sample selection, domain adaptation, and ordinal classification to improve pLTV accuracy and address negative transfer.
This paper introduces RLLR, a label‑sensitive reward reinforcement learning method that improves natural language understanding tasks by aligning training objectives with label accuracy, and demonstrates its effectiveness across eight public NLU datasets and real‑world advertising feature evaluation, outperforming standard RLHF and SFT baselines.
This article presents JD Advertising's technical practices for improving GPU throughput and reducing latency in large‑scale recommendation scenarios, covering system challenges, storage and compute optimizations for training, low‑latency inference techniques, and compiler extensions to handle massive sparse models.
This talk explores how 360's advertising recommendation system leverages language models—from BERT to large‑scale LLMs—to improve user interest modeling, feature extraction, and conversion‑rate prediction, detailing practical challenges, engineering solutions, experimental results, and future research directions.
This article summarizes Tencent Advertising's recent research on recommendation models, covering comprehensive feature encoding techniques, solutions to embedding dimensional collapse through multi‑embedding paradigms, and novel methods such as STEM and AME to disentangle conflicting user interests across multiple tasks.
The article explains how NeurIPS ranks among top AI conferences, introduces the newly selected “Auto‑Bidding in Large‑Scale Auctions” competition, outlines its technical background, four generations of bidding strategies—from classic control to generative models—and details the competition’s tracks, rewards, and how researchers can participate.
In real‑time bidding advertising, a multi‑path recall framework quickly filters billions of ads using parallel non‑personalized and personalized strategies—such as hot‑item rules, collaborative‑filtering, skip‑gram vectors, and GraphSAGE embeddings—while respecting targeting constraints, before a ranking stage optimizes eCPM, with effectiveness measured offline and online and future extensions planned with large language models.
The paper introduces CausalMMM, a variational inference framework that integrates Granger causality and graph neural networks to automatically discover heterogeneous causal structures in marketing mix modeling, enabling more accurate GMV prediction and actionable insights for diverse advertisers.
Alibaba’s Mama technical team secured four paper acceptances at the prestigious KDD 2024 conference in Barcelona, presenting advances such as a diffusion‑based generative bidding model, truthful combinatorial bandit mechanisms for two‑stage ad auctions, bi‑objective contract allocation for guaranteed delivery advertising, and a fast local‑search algorithm for complex contract constraints.
The article details how JD's advertising retrieval platform tackles the core challenge of balancing limited compute resources with massive data by optimizing compute allocation, improving model scoring efficiency, and enhancing iteration speed through distributed execution graphs, adaptive algorithms, and platform‑level infrastructure improvements.
This article details Alipay's advertising algorithm upgrades—including sample‑enhanced cold‑start mitigation, cross‑scene and user‑segmented knowledge transfer, and real‑time feature and online‑learning optimizations—that collectively lifted CTR, CPM, and overall business revenue.
The advertising engine uses a five‑stage funnel—retrieval, recall, coarse ranking, fine ranking, and re‑ranking—each optimized with specialized indexes, multi‑channel recall, multi‑objective twin‑tower models, deep CTR/CVR predictors, and cold‑start paths, delivering up to 33 % spend growth, 6 % eCPM lift and lower latency while maintaining diversity.
This article presents JD Advertising's 2023 innovations that combine relation‑aware diffusion models, category‑aware background generation, and planning‑and‑rendering pipelines to automatically produce high‑quality, scalable, and personalized e‑commerce ad posters, addressing efficiency, cost, and creative limitations of manual design.
The article presents a comprehensive overview of JD's advertising retrieval platform, detailing how it balances limited compute resources with massive data through adaptive compute allocation, distributed execution graphs, elastic systems, and multi‑stage algorithmic improvements to achieve high‑performance, scalable ad matching.
This article presents Baidu's exploration of multimodal content understanding for commercial advertising, detailing the ViCAN pre‑training model, its contrastive and mask‑language learning tasks, integration across recall, ranking and risk‑control pipelines, quantization with MMDict, and future AIGC‑driven generation, all backed by extensive experiments and Q&A.
The article analyzes JD.com's recommendation advertising ranking auction mechanism, detailing its objectives, challenges in traffic value estimation, user interest exploration, and multi‑item auction fairness, and describing the technical evolution from traditional auctions to deep‑learning‑driven solutions.
SizeCube leverages Stable Diffusion‑based diffusion models and a sophisticated pipeline—including quality filtering, feature mining, latent‑space UNet denoising, super‑resolution, and temporal 3D‑U‑Net video processing—to automatically outpaint images and videos to any size, boosting Alibaba advertisers’ creative flexibility, click‑through rates, and asset adaptability across diverse ad placements.
This article presents OPPO's comprehensive advertising recall system, detailing the transition from the old to the new architecture with ANN support, the selection of main‑road recall models, the construction of offline evaluation metrics, sample optimization techniques, model enhancements, multi‑scenario training strategies, and outlook for future improvements.
iQIYI’s ad ranking system replaces static, hash‑based embeddings with TFRA dynamic embeddings to efficiently handle massive sparse ID features, eliminates collisions and I/O bottlenecks, isolates memory during hot model swaps, enabling billion‑parameter models that boost revenue by 4.3 % while planning adaptive embedding sizes for future improvements.
The article details JD Retail Advertising's three‑stage evolution of its online model system—deep‑learning era, large‑model era, and power‑collaboration era—highlighting heterogeneous computing optimizations, platform and system capabilities, distributed graph computing, online learning, and dynamic power allocation to dramatically improve algorithm iteration speed and model performance.
A solo founder built JobBoardSearch.com on a $50‑per‑month VPS, aggregated 400+ job boards, added paid listings and ads, and turned the low‑cost site into a $23,000 profit venture without spending on advertising.
This article presents Baidu's commercial multimodal understanding and AIGC innovations, detailing rich‑media multimodal perception, a unified large‑scale representation framework, scenario‑specific fine‑tuning, and practical applications such as marketing copy, digital‑human video, and poster generation.
Ximalaya’s deep advertising conversion optimization advances from shallow to deep billing models by integrating OCPC dual‑bidding, full‑channel data assistance, and real‑time crowd premium to overcome data sparsity, long conversion delays, and cold‑start challenges, boosting advertisers’ ROI while managing platform risk and guiding future ROI‑protected bidding.
OPPO revamped its advertising recall system by replacing a latency‑prone directional pipeline with an ANN‑based full‑ad personalized architecture, employing a dual‑tower LTR model, multi‑path auxiliary branches, refined offline metrics, price‑sensitive and hard‑negative sampling, and hybrid joint training, which together boosted ARPU by about 15%.
AdvertisingOffline EvaluationRecommendation System
This article presents OPPO's advertising recall system, detailing the transition from the legacy architecture to a new ANN‑based design, model selection criteria, offline evaluation metrics, sample optimization techniques, and various model improvements that together achieved significant ARPU gains.
This article describes the background, problems, and a series of architectural upgrades—including MQ replacement, thread‑pool isolation, Redis/TiKV redundancy, and Spark‑based compensation—to enhance the stability, scalability, and high‑availability of an advertising billing system.
This article presents Huawei's research on applying multi‑task and multi‑scenario learning to advertising recommendation, covering background challenges, algorithm taxonomy, detailed designs of TAML, CausalInt and DFFM models, experimental results, and a Q&A session.
An in-depth look at the advertising conversion funnel explains each stage, compares CPM, CPC, CPI and CPA billing models, outlines how platforms use look-alike algorithms, and details the logic of selecting timely in-app callback events and related-behavior analysis to boost conversion rates.
This article details the business workflow, system architecture, logical index structures, creation, modification, and search processes of the Car Smart Investment advertising index, and evaluates its performance, demonstrating how billions of ad records are efficiently retrieved within milliseconds.
The article explains the architecture of an online advertising system, detailing the roles of ad placement and playback, the design of coarse ranking modules, budget pacing mechanisms, and pricing algorithms that aim to maximize ROI while controlling costs.
The article traces the evolution of effectiveness‑advertising bidding—from simple single‑goal bids to weighted, Pareto‑optimal, and finally adaptive dual‑bidding models that integrate deep‑conversion estimators and a non‑linear control function, enabling platforms to balance shallow cost compliance with deep‑level outcomes such as retention and ROI.
This article explains how Volcano Engine identified the tangled authorization, data‑fetching, and performance problems of its advertising AB‑testing platform and refactored it by splitting services, redesigning the data model with MySQL and ClickHouse, applying DAG scheduling, time‑wheel algorithms, Domain‑Driven Design, and rigorous unit testing to achieve a more stable, extensible backend solution.
This article presents a comprehensive overview of multi‑task and multi‑scenario algorithms applied to recommendation systems, covering background challenges, algorithm taxonomy, recent research, detailed model architectures such as TAML, CausalInt and DFFM, experimental results on public and private datasets, and a Q&A discussion.
This article explains how LinkedIn’s advertising teams prevent misleading AB‑test results, describes the challenges of large‑scale ad experiments such as cannibalization, reviews industry solutions, and introduces their innovative budget‑splitting experiment that dramatically improves statistical power.
By replacing the costly Lambda architecture with a unified data‑lake built on Iceberg and Flink CDC, the advertising team achieved minute‑level latency, strong consistency, and lower storage expenses, cutting end‑to‑end processing times from hours to a few minutes across budgeting, warehousing, OLAP and ETL workloads.
To eliminate data fragmentation, feature inconsistencies, and multi‑language implementation challenges, Bilibili built a unified real‑time advertising feature platform that aligns offline, hourly, and online pipelines via a shared C++ library and JNI, boosting CTR prediction accuracy, cutting training costs, and increasing ad revenue by over 1 %.
AdaSparse introduces an adaptive sparse network that learns a dedicated sub‑network for each advertising scenario, balancing shared and specific knowledge while keeping computational cost low, and achieves +4.63% CTR and -3.82% CPC improvements in Alibaba’s external ad system, as validated on both public and massive production datasets.
Alibaba Health’s user algorithm leverages multi‑channel serialized ad placement, using vector‑based three‑tower models, knowledge distillation, and ROI‑oriented optimizations to sequence user touchpoints, improve conversion rates, and enhance model accuracy across diverse marketing channels.
This article details Autohome's transition from an hourly offline ad data warehouse to a Flink‑StarRocks real‑time architecture, covering background, engine and storage selection, multi‑layer design, implementation steps, encountered issues, monitoring strategies, and future roadmap to achieve second‑level data freshness and high accuracy.
This article presents a data‑science‑focused guide on using causal inference and uplift models to drive overseas ad targeting and user‑operation decisions in games, covering audience selection, privacy‑aware exposure correction, bid optimization, experiment design pitfalls, network effects, and practical recommendations.
This article discusses experimental design challenges in two-sided markets, particularly in advertising scenarios, and presents various methods including four-table experiments, counterfactual interleaving, and contingency table joint sampling to address issues like network effects and competition between supply and demand sides.
This article presents a comprehensive study on insurance creative recommendation, introducing an event‑aware graph extractor, a heterogeneous graph construction, and an adaptive clustering‑gain network that together address data sparsity, counterfactual samples, and cross‑industry cold‑start challenges, achieving significant AUC improvements in experiments.
This article explains the challenges of A/B testing in two‑sided advertising markets and presents several experimental designs—including four‑cell traffic experiments, counterfactual interleaving, joint sampling, and simulation systems—illustrated with Tencent’s practical implementations to mitigate interference, spillover, and competition effects.
The article presents a comprehensive technical roadmap for Alibaba Mama's display advertising cascade ranking system, introducing full‑chain linkage, precise‑value estimation models (PDM, ESDM) and set‑selection approaches (LDM, LBDM), and demonstrates how these innovations jointly improve CTR and RPM while outlining future research directions.
Xiaohongshu’s end‑to‑end intelligent creative platform extracts high‑quality images, generates diverse titles with RED‑pretrained GPT‑2/T5 models, and selects the best ads using a UCB‑based multi‑armed bandit that balances CTR uplift, revenue and user‑experience, while employing position‑corrected metrics and a scalable dual‑tower DNN to boost long‑tail performance and overall revenue.
This article compares advertising and recommendation algorithms, highlighting distinct optimization goals, model design focuses, training methods, implementation principles, auxiliary strategies, and model characteristics, emphasizing how ads aim to increase revenue while recommendations prioritize user engagement and diversity.
The article explains how ad traffic fraud—ranging from simulated impressions to click farms—can be combated using a four‑layer risk‑control system that leverages unsupervised (DBSCAN, Isolation Forest) and supervised (Logistic Regression, Random Forest) algorithms, detailing data pipelines, model training, monitoring, and real‑world case studies.
The article argues that algorithm engineers should move beyond a narrow focus on deep‑learning models, emphasizing the importance of system architecture, data quality, and thoughtful problem framing to break through performance plateaus in advertising and recommendation systems.
This article explains how federated learning is applied to the advertising industry, covering business background, conversion processes from user, client, and server perspectives, algorithmic components such as CTR and CVR models, vertical and horizontal federated learning architectures, compression techniques, and security challenges with corresponding defenses.
This article presents a comprehensive study on estimating video ad attractiveness by analyzing 3‑second completion rates, proposing pairwise MLP and DeepFM models, introducing hierarchical sampling and multimodal features, and demonstrating practical deployment improvements in material recommendation and ad ranking.
Xiaohongshu’s All‑Site Smart投 platform unifies cross‑scene advertising by using AI‑driven keyword extraction, graph‑based user modeling, OCR‑enhanced sparse‑ad handling, learning‑to‑rank targeting, dynamic budget reallocation, and a primal‑dual linear‑programming bidding engine that jointly optimizes ROI under multiple constraints.
This paper proposes a unified guaranteed impression allocation framework (UGA) that jointly models and optimizes contract and real‑time bidding ads, formulates the problem as a non‑convex QCQP, and demonstrates through offline and online experiments that UGA significantly improves platform and advertiser revenue compared to baseline methods.
Vivo tackles the complex problem of mixing heterogeneous content queues—ads, games, and organic items—in its information‑flow and app‑store by evolving from rule‑based weighting to Q‑learning and deep‑learning position models that respect product constraints, preserve ordering, and balance short‑term revenue with long‑term user experience, while planning deeper personalization and on‑device solutions.
The article surveys automatic advertising copy generation, detailing why optimization is needed, the fundamentals of neural text generation with Seq2Seq and attention, extractive versus abstractive approaches, modern embeddings and MASS pre‑training, practical data and evaluation methods, and future enhancements such as multi‑stage attention, knowledge integration, and large pre‑trained models.
This article provides an overview of search, recommendation, and advertising as core internet services, comparing their problem definitions, business goals, algorithmic models, and system architectures across web, e‑commerce, and O2O scenarios, while outlining historical development and key industry examples.
This article describes the design, implementation, and performance of ShouQianBa's advertising billing system, detailing the migration from Apache Druid to Kafka Streams, the architecture for real‑time event processing, data aggregation, persistence, fault tolerance, and the achieved low‑latency, high‑throughput metrics.
The paper introduces EXTR, a Transformer‑based CTR prediction model that jointly encodes diverse externalities from surrounding organic results and ads and infers missing ad placements via a Potential Allocation Generator, achieving superior AUC, COPC and LogLoss on Taobao data and deployment in Alibaba’s advertising system.
This article presents a comprehensive study of the OCPC advertising bidding product, detailing its business logic, system architecture, linear programming formulation, solution methods using GLPK and Gurobi, parameter analysis, PID feedback control, and both offline and online deployment processes.
At KDD 2022, Tencent Advertising introduced the Multi‑Virtual‑Kernel Experts (MVKE) model, a novel multi‑task architecture that replaces traditional dual‑tower structures with virtual‑kernel experts and a gating mechanism to jointly learn user preferences across multiple actions and topics, achieving superior offline and online performance.
Tencent's IEG Growth Middle Platform team announced that two of its research papers—one presenting an actor‑critic reinforcement learning model for real‑time bidding in online display advertising and the other proposing an adversarial adaptation framework for cross‑domain recommendation—were accepted at the top‑tier CIKM conference, highlighting novel algorithms that achieve state‑of‑the‑art performance and have been deployed to serve billions of daily impressions.