0 views collected around this technical thread.
This article examines how systematic data analysis and targeted design experiments—covering color, copy, benefit points, button style, and seasonal skins—significantly increased click‑through rates for app pop‑up resources, offering actionable insights for product managers seeking data‑driven UX improvements.
The paper introduces StEN, a spatiotemporal-enhanced network for CTR prediction in location-based services, combining static spatiotemporal feature activation, dynamic preference activation, and target attention, achieving state-of-the-art offline results and a 1.6% CTR lift in online tests.
This article explains the architecture of video search engine ranking, covering optimization objectives such as relevance, click‑through rate and watch time, and detailing pointwise, pairwise and listwise learning approaches, model training pipelines, and online serving strategies.
This article reviews the evolution of click‑through rate (CTR) prediction models from early logistic regression and factorization machines to deep neural networks, introduces the data‑interaction based RIM (Retrieval & Interaction Machine) architecture with its search and prediction modules, and presents extensive experimental comparisons and future research directions.
The article describes a comprehensive full‑link consistency testing framework for click‑through‑rate models, defining consistency issues, outlining data and logic consistency goals, and presenting a multi‑stage technical solution—including online data capture, offline data stitching, q‑value comparison, and reporting—to ensure model stability and performance.
This article introduces the concept of calibration in trustworthy machine learning, explains why accurate probability estimates are crucial for online advertising, reviews related research and evaluation metrics, and details the evolution of calibration algorithms such as Smoothed Isotonic Regression, Bayes‑SIR, real‑time optimizations, and post‑click conversion models, concluding with engineering deployment and future directions.
This article presents iQIYI's effect advertising system, detailing its dual-engine resource layout, oCPX billing model, algorithmic challenges of high‑dimensional sparse conversion data, the multi‑stage personalized recommendation pipeline, eCPM‑based ranking, online training/inference workflow, and intelligent bidding strategies that balance cost control and traffic quality.
This article reviews recent academic and industrial advances in click‑through rate prediction, classifying optimization techniques for the three‑layer CTR architecture—Embedding, Hidden, and Output—while summarizing three SIGIR papers on graph‑based user behavior modeling, explicit semantic cross‑feature learning, and learnable feature selection for pre‑ranking.
This article explains the theoretical foundations of CTR modeling, why click‑through rates are intrinsically unpredictable at the micro level, the simplifying assumptions that make binary classification feasible, and how evaluation metrics like AUC, contradictory samples, theoretical AUC bounds, and calibration affect model performance.
CTR modeling predicts click probabilities despite inherent microscopic randomness, treating each impression as an i.i.d. Bernoulli event and framing the task as binary classification; because data are noisy and imbalanced, evaluation relies on AUC rather than accuracy, with theoretical upper bounds set by feature quality, and calibration is needed to align predicted values with observed frequencies.
This paper proposes a Path‑based Deep Network (PDN) that combines similarity‑index and embedding‑based retrieval paradigms to model user‑item interactions via Trigger Net and Similarity Net, achieving significant improvements in click‑through rate, GMV, and diversity on Taobao’s homepage feed.
This article systematically reviews the evolution of click‑through‑rate (CTR) prediction models—from early distributed linear models like logistic regression, through automated feature engineering with GBDT+LR, various factorization‑machine variants, embedding‑MLP shallow modifications, dual‑tower combinations, and advanced explicit feature‑cross networks—highlighting each model’s structure, advantages, limitations, and comparative insights.
This article presents a Bayesian smoothing approach to alleviate cold-start problems in click-through rate estimation, introduces key-value memory networks to incorporate prior knowledge, and proposes methods to convert continuous features into dictionary embeddings for deep learning models in recommendation systems.
This article presents iQIYI's effect advertising system, detailing its dual‑engine resource slots, oCPX billing model, algorithmic challenges of high‑dimensional sparse conversion data, the personalized recommendation pipeline, feature engineering across real‑time, short‑term and long‑term signals, and the intelligent bidding mechanism that balances cost control with traffic quality.
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.
This article describes the design and implementation of a comprehensive efficiency‑optimization platform at 58.com, detailing its end‑to‑end workflow—from log aggregation and feature extraction through machine learning model training and online experimentation—highlighting modular, configurable, and scalable solutions for multi‑business, multi‑product ranking.
This article details how 58.com improved the efficiency of its search list ranking by moving from simple time‑based ordering to a comprehensive ranking framework that incorporates feedback strategies, basic machine‑learning models, feature upgrades, and advanced model upgrades, achieving significant gains in click‑through, conversion, and revenue across multiple business lines.
The paper presents “Xianzhi,” a tag‑based precise advertising recommendation algorithm for low‑traffic channels that combines artist‑created material tags, cookie‑derived user tags, and confidence‑adjusted tag‑CTR matrices—enhanced by Wilson intervals, time‑window weighting, dimension aggregation, and objective weighting—to alleviate data sparsity and cold‑start issues, achieving roughly a 10 % CTR lift in online A/B tests.