Tagged articles
recommender systems
54 articles · Page 1 of 1
This paper introduces GRC, a generation-reflection-correction framework for generative recommendation that enables models to self-diagnose and correct full draft semantic ID sequences via structured reflection labels, GRPO reinforcement learning, and entropy-guided scheduling, achieving state-of-the-art results on public and industrial datasets with minimal latency overhead.
CIKM 2026GRPOReflection
0 likes · 12 min read

The article applies recommender system concepts — content pool, feature tagging, indexing, recall, and ranking — to personal knowledge management, advising readers to annotate saved resources with specific problem contexts, build a searchable index, and use AI for summarization and retrieval so that collected materials become actionable when needed.
AI-assisted learningProductivityinformation retrieval
0 likes · 8 min read

Airbnb engineers detail how they adapted search ranking for map interfaces by modeling user attention through three iterations — uniform distribution, tiered markers, and decaying attention from map center — validated via A/B tests that increased bookings by 0.27% and reduced map interactions by 1.5%.
A/B testingAirbnbKDD 2024
0 likes · 10 min read

This article systematically dissects the re-ranking stage in recommender systems, explaining how hard constraints (filtering, quotas, safety) and soft optimizations (diversity, blending) are orchestrated in a prioritized pipeline to transform ranked candidates into a final display list that balances user experience, platform health, and operational needs.
MMRRe-rankingdiversity
0 likes · 40 min read

This article details the multi-path recall architecture in recommender systems, covering five recall sources (tag, vector, popularity, collaborative filtering, exploration), the Recall Manager's parallel scheduling and circuit breaking, engineering deployment pipelines with offline/online evaluation, and future trends like generative recall and joint optimization.
A/B testingANN vector searchRecall Manager
0 likes · 38 min read

AgentX introduces an agent‑driven closed‑loop that automates idea generation, code production, online A/B testing, and experience consolidation, boosting experiment concurrency eight‑fold, increasing per‑person business value 3.7×, and delivering over 0.5% app‑time growth and more than 1 billion RMB annual revenue.
A/B testingAgentXSelf-Iteration
0 likes · 13 min read

The paper introduces the Deep Research paradigm and the RecPilot multi‑agent framework, which transform traditional list‑based recommender systems into proactive decision‑support assistants that simulate user exploration, generate structured reports, and demonstrably outperform existing baselines on TMALL data.
Deep ResearchLLMRecPilot
0 likes · 10 min read

The article analyzes a recent paper that introduces a generative reasoning re‑ranker for LLM‑driven recommendation, detailing its SFT and RL training pipeline, semantic‑ID embedding, target vs. reject sampling strategies, and experimental gains of 2.4% Recall@5 and 1.3% NDCG@5 over the OneRec‑Think baseline.
Generative ReasoningLLMRe-ranking
0 likes · 9 min read

The paper introduces Deep Research for Recommender Systems, a multi‑agent framework called RecPilot that replaces traditional list‑based recommendations with AI‑driven exploration, trajectory simulation, and structured decision‑support reports, and demonstrates its superiority on TMALL data through extensive trajectory and report‑generation evaluations.
Deep ResearchLLMRecPilot
0 likes · 10 min read

DiffNBR introduces a dual‑path diffusion framework combined with an information‑bottleneck mechanism to jointly model spatial co‑occurrence and temporal evolution in next‑basket recommendation, achieving state‑of‑the‑art performance and effectively disentangling repetitive and exploratory purchase patterns.
DiffNBRdiffusion modelinformation bottleneck
0 likes · 8 min read

This article traces the three‑decade evolution of recommender systems—from early collaborative‑filtering prototypes like Tapestry, through the Netflix Prize era and deep‑learning breakthroughs such as Wide&Deep and DIN, to the current generative‑AI wave driven by large language models—highlighting key milestones, technical shifts, industrial deployments, and future challenges.
collaborative filteringdeep learningindustrial deployment
0 likes · 38 min read

CoderRec, a new sequential recommendation framework jointly developed by Tencent Advertising Technology and Tsinghua University, combines domain‑specific latent reasoning with cross‑scale model collaboration to capture implicit user intent and fuse large‑language‑model semantics with traditional recommender signals, achieving state‑of‑the‑art performance on multiple Amazon datasets.
artificial intelligencecross-scale collaborationlarge language models
0 likes · 17 min read


NewBeeNLPOct 13, 2024 · Artificial Intelligence
This article provides a thorough review of negative sampling techniques in recommender systems, categorizing existing methods into five groups, detailing their sub‑strategies, advantages, challenges, and future research directions to improve model accuracy and robustness.
AInegative samplingrecommender systems
0 likes · 11 min read

This article introduces causal inference concepts, explains structural causal and potential‑outcome frameworks, presents the MACR model for debiasing popularity in recommender systems, and details two experiments conducted on the ZhaiZhai platform along with future research directions.
Causal InferenceMACRcounterfactual reasoning
0 likes · 13 min read

This article surveys recent advances in applying causal inference to recommender systems, presenting three lines of work—causal embedding for interest‑conformity disentanglement, contrastive learning for long‑term and short‑term interest separation, and adversarial debiasing of duration bias in short‑video recommendation—along with experimental validation and insights.
Causal Inferencebias mitigationinterest disentanglement
0 likes · 24 min read

This article examines the role of reranking in modern recommendation pipelines, explains why context‑aware listwise models are needed, surveys the evolution from pointwise to generative and diversity‑aware approaches, and reviews real‑world deployments at companies such as Kuaishou, Alibaba, WeChat, iQIYI, and Meituan, highlighting key challenges, evaluation metrics, and business‑rule integrations.
Rerankingdiversityindustry practice
0 likes · 28 min read

The presentation surveys recent causal‑inference research for recommendation systems, introducing the DICE framework to separate user interest from conformity, the CLSR model to disentangle long‑term and short‑term preferences, and the DVR approach with WTG metrics to debias short‑video recommendations, demonstrating improved accuracy, fairness, and interpretability.
Causal Inferencebias mitigationinterest disentanglement
0 likes · 23 min read

This presentation reviews recent research on applying causal inference to recommendation systems, covering causal embedding for separating user interest and conformity, contrastive learning for disentangling long‑term and short‑term interests, and a debiasing framework for short‑video recommendation that uses watch‑time‑gain metrics and adversarial learning to mitigate duration bias.
Causal Inferencebias mitigationinterest disentanglement
0 likes · 23 min read


ArchitectJan 12, 2024 · Artificial Intelligence
The article reviews the arXiv paper “Divide and Conquer: Towards Better Embedding‑based Retrieval for Recommender Systems from a Multi‑task Perspective”, explaining how grouping candidates, balancing easy and hard negatives, and using multi‑interest user vectors can improve recall performance in large‑scale recommendation pipelines.
Divide and Conquerembedding retrievalindustry insights
0 likes · 7 min read

This article systematically reviews the coarse‑ranking stage of recommendation systems, comparing it with recall and fine‑ranking, defining evaluation metrics, detailing sample design, presenting two technical routes, and exploring optimization directions such as dual‑tower models, knowledge distillation, lightweight fully‑connected layers, multi‑objective and multi‑scenario modeling, followed by practical case studies and results.
Evaluation Metricscoarse rankingdual-tower
0 likes · 22 min read

This article explores how large language models can be incorporated into recommender systems, discussing background challenges, specific integration points across the recommendation pipeline, practical implementation methods, experimental results, and future research directions, while highlighting industrial considerations and potential improvements.
LLMfeature engineeringindustrial applications
0 likes · 20 min read

KuaiSim is a comprehensive user simulation environment for recommendation systems that models immediate, long‑term, and cross‑session feedback, supports list‑wise, whole‑session, and retention tasks, provides baselines and evaluation metrics, and demonstrates superior performance on KuaiRand and ML‑1M datasets.
KuaiSimUser Simulationbenchmark
0 likes · 14 min read

The article surveys sequential recommendation models from early non-deep approaches like FPMC, through RNN-based GRU4Rec and CNN-based Caser, to Transformer-based methods such as SASRec, BERT4Rec, TiSASRec, and recent contrastive-learning techniques, recommending SASRec or its variants for production use.
Contrastive LearningTransformerdeep learning
0 likes · 17 min read

The article highlights five CIKM 2023 papers covering a lightweight model‑compression framework for recommender systems, a query‑dominant user‑interest network for large‑scale search ranking, a causal watch‑time labeling approach for short‑video recommendation, implicit negative‑feedback optimization for short‑video feeds, and the KuaiSAR unified search‑and‑recommendation dataset, each with download links, author lists, and key findings.
DatasetKuaishoumodel compression
0 likes · 12 min read

This article surveys how large language models can be incorporated into recommender systems, discussing their strengths and limitations, outlining where and how they can be applied across the recommendation pipeline, presenting recent research examples, and highlighting challenges and future directions for industrial deployment.
LLMfeature engineeringrecommender systems
0 likes · 20 min read

This paper introduces GFN4Rec, a generative flow network approach for listwise recommendation that models the entire list generation as a probability flow, optimizing list-level reward to simultaneously improve recommendation accuracy and diversity, and validates its effectiveness on multiple datasets and simulators.
AIGFlowNetGenerative Models
0 likes · 8 min read

The paper introduces a Generator‑Evaluator (GE) architecture that end‑to‑end re‑ranks information‑flow items using a pointer‑network seq2seq generator and a reward‑estimating evaluator, jointly optimizing relevance and business utilities such as diversity, traffic control, inter‑group ordering, and fixed‑slot insertion, achieving over 70% better‑percentage and significant online gains on Taobao.
Information FlowRankinggenerator-evaluator
0 likes · 19 min read

This presentation introduces a decoupled domain‑adaptation network that separates popularity and attribute representations to mitigate popularity bias in recommender systems, describing the problem, existing IPS and causal‑inference solutions, the CD2AN architecture, experimental results, and practical Q&A.
AIdomain adaptationmachine learning
0 likes · 13 min read

The paper proposes a dual‑interest decomposition head‑attention model that uses a feedback‑aware encoding layer, a factorized head attention mechanism, and separate positive/negative interest towers to improve sequence recommendation performance on short‑video and e‑commerce datasets.
AIFeedbackSequence Modeling
0 likes · 8 min read

This article provides a comprehensive overview of explainable recommendation, covering its motivation, evolution, graph‑based structured explanation techniques, natural‑language generation methods, recent research advances, and open challenges such as fact‑checking, low‑resource scenarios, and evaluation metrics.
Natural Language Generationexplainable recommendationrecommender systems
0 likes · 17 min read

The article explores how to recommend visually appealing game cosmetics—such as character outfits and weapon skins—by transforming subjective notions of beauty into objective features using popularity heuristics, tag‑based labeling, and latent factor models to predict player preferences.
Personalizationgame cosmeticslatent factor model
0 likes · 8 min read

This article examines the shortcomings of current recommender system benchmarks, explains why standardized datasets and metrics are essential, and highlights key findings from the recent BARS paper that propose a more open and reproducible benchmarking framework for recommendation research.
AIBARSbenchmarking
0 likes · 6 min read

This article reviews the emerging field of explainable recommendation, covering its motivation, historical evolution from template‑based to knowledge‑graph and generative‑language approaches, recent advances in graph‑structured and natural‑language explanations, key research works, industrial applications, and open challenges such as fact‑checking, low‑resource settings, and evaluation methods.
AINatural Language Generationexplainable recommendation
0 likes · 16 min read

PICASSO, Alibaba’s GPU‑centric sparse training engine for wide‑and‑deep recommender systems, merges identical embedding tables, interleaves data and kernel operations, and caches hot embeddings on GPU, eliminating the parameter server and delivering up to tenfold speedups over TensorFlow‑PS while maintaining model quality.
AlibabaGPU optimizationmachine learning
0 likes · 14 min read

This paper introduces MAIL, a model‑agnostic dual‑tower framework that uses a zero‑shot learning tower to generate virtual user behaviors for new users and an embedding‑based ranking tower, achieving 13‑15% CTR lift in large‑scale live‑stream recommendation at NetEase Cloud Music.
EmbeddingZero-Shot Learningcold start
0 likes · 33 min read

This article explains why cold‑start is a critical challenge for recommender systems, outlines four practical strategies—generalization, fast data collection, transfer learning, and few‑shot learning—and then details the DropoutNet model, its end‑to‑end training, loss functions, negative‑sampling techniques, and open‑source implementation.
DropoutNetEmbeddingFew-Shot Learning
0 likes · 21 min read

The paper presents a generative re‑ranking framework for Taobao’s home‑decor channel that combines heuristic sequence generation methods (MMR, DPP, beam search) with a context‑aware encoder to produce diverse, relevance‑balanced recommendation lists, achieving notable gains in PV, IPV, CTR and click‑diversity over traditional point‑wise ranking.
context-awarediversitygenerative re-ranking
0 likes · 19 min read

This article reviews algorithmic solutions for cold‑start recommendation, covering the efficient use of side information, knowledge graphs, cross‑domain transfer, multi‑behavior signals, limited interaction data, explore‑exploit tactics, and additional practical scenarios, while summarizing key methods such as DropoutNet, MetaEmbedding, MWUF, MeLU and MetaHIN.
cold-startcross-domainknowledge graph
0 likes · 11 min read

This article reviews how knowledge graphs can be integrated into recommender systems to address data sparsity and cold‑start problems, covering collaborative filtering limitations, KG embeddings (TransE, TransH, TransR), deep knowledge‑aware networks, multi‑task feature learning, RippleNet, KGCN, experimental results, and a comparative analysis of performance, scalability, and interpretability.
Embeddingartificial intelligencecollaborative filtering
0 likes · 11 min read


58 TechSep 7, 2020 · Artificial Intelligence
This article presents a comprehensive study on improving individual diversity in recommendation systems by detailing architectural optimizations across recall, rule, and re‑ranking layers, explaining the principles and practical deployment of MMR and DPP algorithms, and demonstrating their impact on key business metrics through extensive experiments.
Custom DistanceDPPMMR
0 likes · 18 min read

This article reviews music recommendation techniques, focusing on content-based methods using metadata and audio features, and context-aware approaches that incorporate environmental and user-related factors, highlighting challenges, classification of metadata, acoustic descriptors, and integration strategies for personalized music services.
audio featurescontent-basedcontext-aware
0 likes · 21 min read


HomeTechJun 10, 2020 · Artificial Intelligence
This article introduces recommender systems and the exploitation‑exploration dilemma, explains common E&E algorithms such as ε‑greedy, Upper‑Confidence‑Bound, and Thompson Sampling, and details their practical deployment for interest‑point eviction, selection, and adaptive recall count optimization in an automotive recommendation platform.
Bandit AlgorithmsEpsilon-GreedyExploitation
0 likes · 10 min read

This talk presents a real-time attention-based look‑alike model (RALM) designed to address the long‑tail problem in recommendation systems by efficiently expanding seed users, leveraging user representation learning, attention mechanisms, and clustering to deliver timely, diverse content without retraining the model.
AttentionLong TailUser Representation
0 likes · 24 min read

This article reviews recent advances in applying reinforcement learning to recommendation systems, explains the fundamental RL concepts, discusses the specific challenges such as large action spaces, bias, and long‑term reward modeling, and summarizes two influential YouTube papers along with practical insights and future directions.
Off-PolicyUser Modelinglong-term reward
0 likes · 13 min read

This article surveys traditional and deep learning based collaborative filtering techniques—including similarity methods, matrix factorization, explicit and implicit feedback handling, various loss functions, evaluation metrics, and TensorFlow implementations of GMF, MLP, NeuMF, DMF, and ConvMF models—providing practical guidance for building large‑scale recommender systems.
Evaluation MetricsTensorFlowcollaborative filtering
0 likes · 21 min read

This article presents a comprehensive overview of Club Factory's recommendation system, covering its product recommendations, key challenges such as user and item cold‑start, the modular architecture, detailed recall and ranking processes, and practical considerations for deployment in e‑commerce.
AIRankingRecall
0 likes · 3 min read


21CTOOct 25, 2018 · Artificial Intelligence
This article explains what recommender systems are, their evolution, when to use them, the main techniques—including collaborative filtering, content‑based and knowledge‑based approaches—addresses cold‑start challenges, and provides a step‑by‑step Python implementation with code examples.
AIPythoncollaborative filtering
0 likes · 15 min read


JD TechFeb 1, 2018 · Artificial Intelligence
The Telepath model, presented at AAAI 2018, leverages a biologically‑inspired visual extraction pipeline and dual interest‑understanding networks to improve ranking in large‑scale e‑commerce recommendation and advertising, achieving significant offline and online gains in CTR, GMV, and ROI.
AAAI 2018Telepathdeep learning
0 likes · 13 min read

This article explains why recommender systems were created, describes the problem of information overload, introduces core algorithms such as popularity, content‑based, collaborative filtering and hybrid methods, and illustrates each with a six‑user/book example and a Netflix case study.
collaborative filteringcontent-based filteringhybrid recommendation
0 likes · 17 min read

This article reviews the early research on applying deep learning techniques such as autoencoders, stacked denoising autoencoders, and hybrid collaborative‑filtering models to recommender systems, describing the underlying matrix‑factorization theory, side‑information integration, experimental results, and future prospects.
Hybrid Modelautoencodercollaborative filtering
0 likes · 13 min read

This article explains the difference between explicit and implicit user feedback, discusses their advantages and pitfalls, and shows how collaborative‑filtering techniques such as user‑based, item‑based, adjusted cosine similarity, and Slope One can be applied to build accurate recommendation engines.
Slope Oneadjusted cosine similaritycollaborative filtering
0 likes · 19 min read


21CTOSep 20, 2016 · Artificial Intelligence
In this talk, Quora’s VP of Engineering Xavier Amatriain shares practical lessons from building the company’s large‑scale recommender system, covering data richness, implicit signals, model choices, feature engineering, evaluation strategies, and why distribution isn’t always required.
Model Evaluationfeature engineeringimplicit feedback
0 likes · 4 min read

This article explains the significance of personalized recommendation, distinguishes it from traditional push services, outlines typical application scenarios, and details a step‑by‑step approach—including user profiling, behavior sampling, algorithm modeling, machine learning, and content lifecycle management—to build effective recommender systems.
Personalized Recommendationinformation overloadrecommender systems
0 likes · 7 min read


21CTOSep 7, 2015 · Artificial Intelligence
This article surveys the fundamental misconceptions about personalized recommendation, distinguishes it from market segmentation and collaborative filtering, and then systematically presents ten critical research challenges—including data sparsity, cold‑start, scalability, diversity‑accuracy trade‑offs, system robustness, user behavior modeling, evaluation metrics, UI/UX, cross‑dimensional data integration, and social recommendation—each illustrated with examples and recent literature.
Evaluation MetricsPersonalized Recommendationcold start
0 likes · 31 min read
