К содержанию
Олег НагорныйRecSys2026

ACM RecSys 2026Миннеаполис · 27.09–02.10обновлено 2026-10-16

Путеводитель по RecSys 2026

Разборы 269 работ конференции, прочитанных целиком: статьи, воркшоп-доклады, доклады симпозиума и tutorial. У каждой — что предложено, какие числа получены и где вывод перестаёт работать.

Открыть каталогОрганизацииВоркшопы и туториалы · 23Облако тегов ↓
Работы по основной теме: 269 · клетка — одна работа, клик открывает разбор · название темы — её каталог
  1. Генеративные рекомендации
  2. Последовательные рекомендации
  3. Языковые модели и агенты
  4. Коллаборативные модели
  5. Оценка и причинность
  6. Объяснения и контроль
  7. Промышленные системы
  8. Кросс-домен и мультимодальность
  9. Реклама и рекомендации

Наведите на клетку или пройдите стрелками — появится название работы.

9направлений
27тематических выводов
264организаций авторов
462связи работа × организация
269/269записей в Zotero
00

Одни и те же работы — три разреза.

Точка — одна работа конференции; меняется только признак, по которому её раскрасили. Разрез 1 — основная тема работы, разрез 2 — организации её авторов, разрез 3 — тип материала. Прокрутите страницу — рисунок перестроится.

  1. 01

    Основная тема

    У каждой из 269 работ одна основная тема — она и задаёт цвет точки. Крупнейшая тема — «Промышленные системы»: 57 работ. Вторичные темы на этот разрез не влияют: они видны на странице работы и в каталоге.

  2. 02

    Организации авторов

    У 80 из 269 работ есть хотя бы один автор из десяти самых частых организаций; остальные 189 распределены по длинному хвосту из 264 организаций. Число у названия — все работы организации. Точка — работа, отнесённая к самой частой из её организаций внутри десятки, поэтому точек в строке может быть меньше.

  3. 03

    Тип материала

    Разбор полного текста есть у всех 269 работ, запись в Zotero — у 269. 251 — публикация конференции; остальное — докторантский симпозиум, воркшоп-доклады, tutorial и отчёт челленджа.

01

От задачи — к работам.

У каждой работы одна основная тема и до двух вторичных тегов. Основная тема задаёт место в списке, вторичные показывают, где направления пересекаются.

  1. 01Генерация и семантические кодыГенеративные модели выдают не готовую выдачу, а код, намерение или структуру страницы. Качество решает не только декодер: важно, что именно закодировано и как код связан с поведением.3737 в каталоге →
  2. 02История, время и последовательностиДлина истории, порядок событий и выразительность энкодера — разные свойства, и работают они по-разному. Прежде чем усложнять модель, надо понять, какое из них использует конкретная задача.6161 в каталоге →
  3. 03Языковые модели, диалог и агентыЯзыковую модель ставят энкодером, планировщиком, памятью или оркестратором. Полезность зависит от роли в системе, а не от размера модели и не от наличия пошаговых рассуждений.8282 в каталоге →
  4. 04Коллаборативность и геометрия представленийГраф, разреженный код и смешанный эмбеддинг по-разному переносят коллективный сигнал. Отдельный вопрос — полезно ли соседство, дёшево ли это обслуживать и что означают координаты.5050 в каталоге →
  5. 05Оценка, причинность и достоверность выводаМетрика отвечает на вопрос только вместе с протоколом. Если изменить пул кандидатов, наблюдаемость или случайность обучения, изменится сам объект сравнения.8686 в каталоге →
  6. 06Контроль, объяснения и интересы участниковРавномерная выдача, понятный текст и редактируемая память ещё не означают справедливости, верного объяснения и реальных полномочий человека.7575 в каталоге →
  7. 07Промышленное ранжирование и стоимостьВыигрыш в проде чаще даёт согласованность компонентов: отбор кандидатов, цели, частота обновления и бюджет вычислений должны сходиться вместе.133133 в каталоге →
  8. 08Домены, группы и мультимодальные задачиВ новом домене меняются не только признаки. Меняются допустимость действия, горизонт пользы, участники решения и смысл доступной обратной связи.7373 в каталоге →
  9. 09Реклама и рекомендацииРекламный показ и органическая рекомендация живут в одной выдаче, но оптимизируются разными целями. Доля клика, ценность конверсии и занятое место измеряются отдельно, а один показ расходует общий ресурс кампаний.1818 в каталоге →
02

Облако: чем больше слово, тем больше работ.

Что показать

2 → 18работ; площадь слова растёт как √n. Клик открывает каталог с этим фильтром. Теги из разборов, у которых 2 и больше работ: 78 из 333; остальные 255 — на страницах работ.

03

Темы редко живут поодиночке.

Это карта нашей разметки, а не рейтинг научной значимости и не проекция эмбеддингов.

Распределение тем

Способ подсчёта

Каждый разбор учтён один раз; сумма равна 269.

Генерация и семантические коды
27
История, время и последовательности
21
Языковые модели, диалог и агенты
26
Коллаборативность и геометрия представлений
12
Оценка, причинность и достоверность вывода
43
Контроль, объяснения и интересы участников
46
Промышленное ранжирование и стоимость
57
Домены, группы и мультимодальные задачи
24
Реклама и рекомендации
13

Матрица совместных тем

Ячейка — число разборов с обоими тегами. Диагональ — все материалы темы. Нажмите для списка.

Пересечения тегов, число разборов
ТемаGENSEQLLMGRAPHEVALHUMANPRODDOMAINADS
GEN
SEQ
LLM
GRAPH
EVAL
HUMAN
PROD
DOMAIN
ADS
Расшифровка обозначений

GEN — Генерация и семантические коды

SEQ — История, время и последовательности

LLM — Языковые модели, диалог и агенты

GRAPH — Коллаборативность и геометрия представлений

EVAL — Оценка, причинность и достоверность вывода

HUMAN — Контроль, объяснения и интересы участников

PROD — Промышленное ранжирование и стоимость

DOMAIN — Домены, группы и мультимодальные задачи

ADS — Реклама и рекомендации

04

Близость: одна точка — один разбор.

Два пространства. «Текст разбора» — проекция UMAP эмбеддингов (Qwen3-Embedding-0.6B: заголовок, суть, результат, ограничение; темы и теги в текст не попадали). «Литература» — граф: ребро между работами, у которых есть общий источник, вес ребра — число общих ссылок, раскладка графовая, сообщества — Leiden. Общие группы в тексте: кластеры эмбеддингов и редакторские темы. Выберите группу — остальные точки уйдут в фон.

Сообщества по литературе · 6 сообществ · 269 разборов · 262 со списком литературы · Leiden, modularity 0.215 · 4362 связей. Нажмите на точку, чтобы открыть разбор.
Группировка
A Control Function Framework for Mitigating Position Bias in Learning to Rank SystemsA Redundancy Reduction Approach for Controllable Sequential RecommendationsAdaptive Retraining of Recommender Systems via Reinforcement LearningAddressing Cross-Stage Decoupling of Semantic and Collaborative Signals in Generative RecommendationAutomated Selection-based Mixture-of-Experts with Dual-stage Input–Target Pattern Learning for Sequential RecommendationBeyond Fixed Depths and Widths: Optimizing Textual Decoding Tries in LLM-based Generative Recommendation (BONSAI)BiLPR: Bidirectional Teacher-Student Agent Interaction for Context-Aware Learning Path RecommendationBilateral Intent-Enhanced Sequential Recommendation with Embedding Perturbation-Based Contrastive LearningBridge the Unseen Gap: Enhancing Non-overlapping Cross-domain CTR Prediction via Profile RetrievalCONGA: Continual Neural Gated Architecture for Long-History Sequential RecommendationCalibrating User Preferences for Cross-Domain Recommendation via Target-Guided Representation MappingCoarse-to-Fine Long-term Interest Modeling for Generative RecommendationConsensus vs. Dissent: Dynamic LLM Modeling of Subjective Preferences in Group RecommendersDart: Adaptive Tweedie Objective Mitigates the σ Escape Route in Cross-Platform Popularity PredictionDP-Rec: Towards Dynamic Patching for Efficient Long-Sequence RecommendationDPGFlow: Decoupled Preference Guided Flow Matching for Cross Domain Sequential RecommendationDPGR: Dual-Domain Spatiotemporal Generative Retrieval for Intent-Aware Local Life Service RecommendationDecoupled Learning and Selection in Slate Recommendation for Privacy and Stability Under Noisy ScoresDo We Care About Personalization and Explainability? An Interview Study with News Recommendation EngineersDo We Really Need LLMs to Augment All? A Selective Augmentation Framework with Lightweight Language Models for Multimodal CTR PredictionEmpowering Cross-Domain Sequential Recommendation with Hybrid Tokenization and Serial-Parallel DecodingResidual Dominance as a Structural Account of Last-Item Reliance in Causal Self-Attention RecommendersFASC: A Feature Aspect-Level Sentiment Consistency Framework for Explainable Recommendation EvaluationFiCoRec: Fine-Grained Contrastive Learning with Dual Aggregation for Sequential RecommendationFrom Constraint to Control: Modeling Expected Fairness in Ranking SystemsGSPRec: On Improving Item Representations in Graph Signal Processing for Collaborative FilteringGTP: Mitigating Popularity Bias in PLM-based Sequential Recommendation via Group-Aware Token PruningGateBoxGCN: Hard-Soft Gated Box Embeddings with Graph Convolution for RecommendationGeneralized Position-Based Model: Rethinking Position Weights in Ranking Off-Policy EvaluationGenerating Personalized Images for Sparse-Interaction Users with Uncertainty-Aware Retrieval and Dense Knowledge GuidanceGive the Long-tail More SPACE: Promoting Provider Fairness in Next POI RecommendationHamiltonian Spectral-Temporal Dissipative Dynamics for Sequential RecommendationCoFiRec: Coarse-to-Fine Tokenization for Generative RecommendationImproving Rare Medication Recommendation with Counterfactual Data Augmentation and Large Language ModelsInformation-Aware Long Sequence Compression for Sequential RecommendationInverse Theory of Mind Modeling for Content Recommendation: From Web Browsing to Dynamic Intelligent InterfacesLearning Sparse Representations of Multimodal Content for Enhanced Cold Item RecommendationMODE: Mutual Optimality in Direct Effects of Reciprocal Recommendations in Matching MarketsMaskPoison: Intent-Guided Poisoning Attacks on Sequential Recommendation via Masked Discrete DiffusionMembership Inference Attacks on In-Context Learning RecommendationNot All Branches Are Equal: Adaptive Semantic ID Construction for Generative RecommendationOn Reranking Space for Multi-Tenant Retrieval with Adapted QueriesOn the Convergent Validity of Offline Evaluation Designs for Recommender SystemsProbabilistic Residual Learning for Online RecommendationsPurifying Interaction Sequences: Topology-Aware Spectral Denoising for Side-Information Integrated Sequential RecommendationRecRec: Latent Interests Recursive Reasoning for Sequential RecommendationReducing Perceived Polarization through Affect-Balanced News ReframingSAGE: Sequence-level Adaptive Gradient Evolution for Generative RecommendationSPEAR: Selection-aware Personalized End-to-end Adaptive Rewriting and Retrieval for Community SearchSafety-Aware Next-POI Recommendation with Large Language ModelsSimulating Diverse User Behavioral Stereotypes for Evaluating Agentic Conversational RecommendersStabilizing Stability and Plasticity in Graph-based Continual Recommender SystemTAGRec: Tailness-Aware Gate for Sequential Recommendation under Long-Tailed DistributionsTRACE: Targeted Ranking-Aware Counterfactual Explanation for Sequential RecommendationTSMOO: Solving Multi-Objective Experimentation with Constrained Thompson SamplingThe Cost of Continuous Time: Diagnosing Solver Sensitivity in Recommenders and Mitigating It via Training-Free RoutingThe Utility of LLMs in Recommender Systems Explanation EvaluationThere’s Something About You: Epistemic Recommendation for Latent Interest DiscoveryTopology-Aware Tokenization for Generative RecommendationTowards welfare-oriented recommendations in activity-travel behaviorUniRec: A Unified Expressive-Aligned Generative Recommendation Framework for E-commerceWhen Do Contrastive Explanations Really Matter in Recommender Systems?When attention is bounded, structure matters: personalizing recommendation explanations under time constraintsZero-Observation User Reactivation with Gap-Driven Dimensional Gating (DeltaGate)A Causal Transformer Multi-Touch Attribution with Dual Debiasing and Explainable Visualization for Advertising RecommendationAssessing Sentiment Semantics in KG-Based Explainable Recommender SystemsBeyond Index-Only RoPE: Integrating Time and Order for Generative RecommendationBookForYou: Leveraging BART-Generated Narrative Tropes for Content-Based Book RecommendationsCandidate Retrieval for Provider Fairness under Severe Recommendation ImbalanceChoosing What Matters: Query-Aware Multimodal Routing for Conversational RecommendationComparability for Recommendation Model Evaluation and Selection from a Multi-stakeholder PerspectiveCovariance-Aware Newton-Schulz Orthogonalization for Noise-Robust Sequential RecommendationDistribution-Level Contrastive Supervision for Generative RecommendationDual Conditional Diffusion for Generative Cross-Domain Recommendation via Disentangled Knowledge TransferEnriching Graph-Based Modeling with Sequential Signals for Explainable Course RecommendationGradSup: Gradient Superposition for Personalised and Scalable LLM RecommendationHierarchical GTV Estimation: Bridging the Gap between Container-Level Ranking and Item-Level ConversionImpression Share Prediction: An Offline Evaluation Task for Ads Ranking SystemsInteraction Modality and Trust: Investigating System-Driven Conversational Recommendation and Faceted SearchLLM-as-a-Judge for Evaluating System Responses in Conversational Music RecommendationPersonalized Recommendation Tool Learning via Autonomous Language AgentsPosition Bias Undermines Preference Consistency in Listwise LLM-Based RerankingPriCoRec: A Privacy-Aware Cloud–Device Collaborative Framework for Ad Recommendation under Feature ConstraintsProbing the LLMs for Generating User Profile of Session-based Social RecommendationScaling LLM-Enhanced Linear Autoencoders to Industrial-Size CatalogsStructure-Preserving Projection for Mitigating Modality Bias in LLM-Based Sequential RecommendationSupport Gap: Selecting Fixed-K Candidate Sets for Retained Personalized HeadroomSurface Matching in Skill Recommenders: Incomplete Evaluation Hides Simplicity BiasTowards Efficient Hyperbolic Representation Learning for Recommender SystemsUncertainty-Aware Gated Context Fusion for Next POI RecommendationExploiting ID-Text Complementarity via Ensembling for Sequential RecommendationWhen Fixed-Candidate Offline Evaluation Changes Model Selection in Two-Stage RecommendersA Position Paper on Recommender Systems in the Era of Autonomous AgentsBackward and Inward: From Preferences to Personalization and the Safety Crisis Within Conversational Recommender SystemsBeyond Passive Preference Inference: User-Governed Preference Memory and Calibrated Initiative in Recommender SystemsCodebook-Based Semantic IDs in Generative Recommendation: Enabling Interface, Emerging BottleneckEachMovie: archaeology of a first latent recommender systemForward and Outward: From Aggregate to Individual and Towards Human Understanding in Conversational RecommendationFrom Side Information and Knowledge Graphs to Large Language Models: Two Decades of Knowledge Integration in Recommender SystemsFrom User–Item Edges to Exposure Allocation: Matching Evidence to Platform-Level Recommender ClaimsSome People Are Worth Recommending For: Twenty Years of Progress, Gaps, and Opportunities For and With Children at RecSysThe Underrated Catalyst: Advancing Critiquing Towards Steerable and Next-Generation LLM Recommendations.Towards the Human-Centered Study of Recommender System ProvidersUser-Controlled Intent Layers for LLM-Mediated Personalization: A Research Agenda for Recommender SystemsWho Are We Recommending To? Recommender Systems in the Agentic WebAttacking and Defending Multi-Agent Collaborative Filtering Systems Through ConnectivityAre We Really Making Progress in Group Recommendation? Unmasking the Tie-Breaking IllusionPersonalised Drug Recommender Systems based on Electronic Health Records: Reproducibility and External ValidityWe’ve Already Been There: A Study of Data Leakage in Ephemeral Group Recommender SystemsBinge Watch: Reproducible Multimodal Benchmark Datasets for Large-Scale Movie Recommendation on MovieLens-10M and 20Mhoploy: a Plugin-Based Inference Layer for Path-Based Explainable Recommendation over Knowledge GraphsJTH: A Dataset for Evaluating Cold-Start and Temporal Dynamics in Job Recommendationscikit-rec: A Unified, Extensible Recommendation Library with scikit-rec-agent, a Conversational Interfaceτ-Rec: A Verifiable Benchmark for Agentic Recommender Systems (τ-Rec)WatchLens: A Configurable Platform for Online Video Recommendation ExperimentsWorkRB: A Community-Driven Evaluation Framework for AI in the Work DomainA Framework for Value-Aligned Modeling and Robust ExperimentationA Self-Triggered Agentic Push Recommendation SystemA Unified Generative Re-ranking Framework with Adaptive Multi-objective FusionAdvancing Relevance Measurement with Vision–Language Models for Web-Scale SearchAlignment + Accuracy: The Cascade Reward Representation for PrerankingAttending to the Core: Core-Task Attention for RecommendationBootstrapping Conversational Recommendation Agents At Spotify: Synthetic Data Generation and Self-Improvement LoopsBreaking the Loop: An Empirical Comparison of Strategies for Novelty and Freshness in YouTube MusicBuilding a User Foundation Model for the Open WebCAPTS: Channel-Aware, Preference-Aligned Trigger Selection for Multi-Channel Item-to-Item RetrievalConAlign: Conditional Alignment Framework for Balancing Biased and Unbiased RecommendationDART - Drift Aware Ranking with forecasTing and uncertaintyDRanker: A Transformer-Based Multi-Task Ranking Model for the Disney+ HomepageEGR: Embedding-Native Generative Retrieval with a Shared LLMEmbedding Subspace Partitioning for Dynamic Multi-Objective RetrievalEntity Representation Learning Through Onsite-Offsite Graph for Pinterest AdsFLUID: From Ephemeral IDs to Multimodal Semantic Codes for Billion-Scale Livestreaming RecommendationFrom Exploitation to Balance: Efficient Retargeting Recommendation via DPO-Guided Slot AllocationFrom M Passes to One: A MapReduce-Style Bootstrap for Pointwise Off-Policy Evaluation in Production RecommendersFrom Placements to Pages: Modeling Recommendation Module Interactions with Sequential Transformer Neural BanditsGenerative Spatiotemporal Intent Sequence Recommendation via Implicit Reasoning in AmapGRIP: Generation and Reasoning for User Profile CompletionGenPage: Towards End-to-End Generative Homepage Construction at NetflixGuess Where You Go: Generative Next Point-of-Interest Recommendation in AmapHIPA-Net: Heterogeneous Intent Projection Alignment for Feedback Calibration in CVR PredictionHeterogeneous Ranking in Industrial-Scale Recommender Systems: A Case StudyHybrid Retrieval and Multimodal Reranking for Item-to-Item Recommendation in a C2C MarketplaceHypothesis-Driven Shelf Generation for Personalised RecommendationIDProxy: Cold-Start CTR Prediction for Ads and Recommendation at Xiaohongshu with Multimodal LLMsInterest Sequence for User Modeling in Industrial Short-Form Video RecommendationLLM Reasoning for Subjective Tasks: Failure Modes, Mitigation, and Dynamic Reasoning RoutingLLM-Based Re-Ranking for Real Estate SearchLLM-Based User Personas for Recommendations at ScaleLO-FAR: A Cost-Aware Local Filter for Sparse Feature Ranking in Industrial Ad RecommendationLearned Cross-Task Relationships in Multi-Task ModelsLong-term User Engagement Optimization through Model-agnostic Downstream Rewards LearningM2R: End-to-End Implicit Scenario Discovery for Multi-Scenario and Multi-Crowd Ranking at TaobaoMDGR: Masked Diffusion Generative RecommendationMESH: Scaling Up Retrieval with Heterogeneous Content UnificationMelo: A Production LLM-Powered Music Recommendation AgentMend the Measurement Gap: Latent User Preference Modeling for Short-Form Video RecommendationMoR: An Adaptive Retrieval Allocation Balancing Long-Term Interest and Short-Term EvidenceMosaic: A Fleet of User Embedding Specialists for Recommendation at MetaMulti-Objective Ranking for Live-Streaming: Balancing Fresh and Delayed Signals with Segment-Aware TargetingMulti-Probe Zero Collision Hash (MPZCH): Mitigating Embedding Collisions and Enhancing Model Freshness in Large-Scale RecommendersMultilingual Semantic Retrieval for Apple Music SearchNextGen: A Multi-Objective Generative Re-ranking Framework for Taobao RecommendationPLAIN: An Explainable Generative Search System Enhanced by Multi-granularity Semantic AlignmentPOEM: Partial-Order Enhanced Real-Time Sequential Modeling for RecommendationPROMISE: Process Reward Models Unlock Test-Time Scaling Laws in Generative RecommendationsPersonalizing Incremental Video Search with Hybrid Text and ID EmbeddingsPinDCO: Whole-Page Aware Dynamic Creative Optimization at ScaleProgressive Alignment of Recommender Foundation Model through Multi-Phase Post-TrainingRECAP: Feedback-Driven Streaming Semantic User Profiles for Short-Video RecommendationREEF: Real-time end-to-end Explore-Exploit framework for e-commerce FeedsRankGraph-2: Lifecycle Co-Design for Billion-Node Graph Learning in RecommendationRecEvolve: A Knowledge-Driven Autonomous Agent System for Recommender SystemsSMART: LLM-Augmented Hybrid Retrieval for Dynamic Product AdsSample Is Feature: Beyond Item-Level, Toward Sample-Level Tokens for Unified Large Recommender ModelsSelf-Evolving Recommendation System: End-To-End Autonomous Model Optimization With LLM AgentsSequenceO1: End-to-End Ultra-Long (100K) Sequence Modeling in Recommendation with Low-Rank CachingShape Your Feed: An LLM-based Agentic System for Conversational RecommendationSoft Curriculum Learning for Optimizing Fresh and Generalized RecommendationsSpace Efficient Item Embeddings for Large Recommender SystemsStructuring and Tokenizing Distributed User Interest Context for Generative RecommendationTSSR-Beta: Enhancing Billion-Scale E-Commerce Semantic Retrieval via Representation-Level InteractionThe Text on the Creative: An Under-Exploited Ranking Modality for Short-Form Video AdsToken Factory: Efficiently Integrating Diverse Signals into Large Recommendation ModelsTokens are All You Need: Dual-purpose Semantic IDs for Achieving LLM-Level I/O Efficiency in Recommendation Systems (SiDec)Towards Full Candidate Interaction: A Comprehensive Comparison Network for Better Route RecommendationTowards Generalizable and Efficient Large-Scale Generative RecommendersTowards Reliable Social A/B Testing: Spillover-Contained Clustering with Robust Post-Experiment AnalysisTransAct V2: Lifelong User Action Sequence Modeling on Pinterest RecommendationTubiFM: Unified Item, Carousel, and Search Ranking for Streaming DiscoveryUNIQUE: A Unified Retrieval and Ranking System for Large-Scale Feed RecommendationUncertainty-Aware Reward Modeling: A Large-Scale Case Study in Video RecommendationUniPinRec: Unifying Generative Retrieval and Ranking at Pinterest ScaleUniShare: A Unified Framework for Joint Video and Receiver Recommendation in Social SharingUniTraj: Cross-Domain Long-Sequence Modeling for Commercial RecommendationVersioned Late Materialization for Ultra-Long Sequence Training in Recommendation Systems at ScaleWHALE: A Scalable Unified Model for Recommendation with Wukong-HSTU ArchitectureWhich LLM to Fine-Tune? Agent-Driven Model Selection at ScaleBlended User Embeddings for Steerable Matrix FactorizationCalibrating Recommendations on Ordinal AttributesCoSID: Concept-Conditioned Semantic-ID Decoding for Efficient Generative RecommendationCross-Spectrum Candidate Set Analysis for Popularity-Aware RecommendationDeadweight Loss in Recommendation SystemsDo SID Alignments Help Generative Recommendation? A Research and Practice Note on OpenOneRec DiagnosticsDo Sequential Recommendation Benchmarks Really Require Higher-Order Sequence Modelling?Don't Waste the Last Token: A Quantizer-Agnostic Quality Boost from Semantic ID Collision ResolutionExploring Revealed Behavior and Stated Profiles in Longitudinal News RecommendationHow Faithful Is the Reasoning of LLM Recommenders? A Counterfactual AuditIn-Batch Negatives Can Silently Cripple LLM-Encoded Sequential RecommendersIntent-Description Anchoring Bias in LLM-as-a-Judge Evaluation of Recommender SystemsMonte Carlo Power Analysis for Small-System A/B Trials: Detecting Differences in Newsletter EngagementMore Test Users, More Overconfidence: Seed Variance in Sequential Recommender TestsMusic Discovery Quality and the Value of FamiliarityPCP-Decoding: Training-Free Personalized Category-Relative Popularity Calibration for Generative RetrievalPairwise Ranking Outperforms Single-Action RL for Offline Explanation Selection: A Practical LessonPersonalized Fashion Discovery via Session Trajectories and Soft NegativesPredicting Custom-Feed Returns for New Bluesky Posts: A Prospective StudyRepetition Is Not Enough: Evaluating Repeat Recommendations Through Utility and TimingRevisiting the Assumptions of Shared-Account Sequential RecommendationSERA: A Multi-Dimensional Evaluation Framework for Retail Agentic Shopping AssistantsScaling Sequence Learning under Production Latency ConstraintsShould I Be Polite to My LLM Relevance Judge? Tone as a Severity Operating-Point ShiftSparse Autoencoders as Semantic Domain Adapters for Recommender SystemsThe Illusion of Control: Commensurability, Concentration, and Correlation in Multi-Objective RecommendationTowards Dynamic Relationship Schema Discovery for Complementary News RecommendationTraining seeds and model-selection stability in recommender-system evaluationWhat price fairness? Evaluating Energy – Fairness – Accuracy Trade-off in Recommender SystemsWhen Should Recommender Systems Not Act?Whose Posts Get Ranked: Identical-Text Exposure Gaps in Bluesky Custom FeedsWhy Do CF and Review Models Disagree? User-Specific Rating–Review Scale GapsYour 5-Core Train Split May Not Be 5-Core: Core Filtering and Temporal Splitting Do Not CommuteAutoRecLab: An Autonomous Recommender Systems LabCOMPRESSO: Espresso-Style Sparse Representation Learning for Interpretable Recommender SystemsCORGI: Communal Feed Governance for BlueskyConversational Recommendation over Live E-Commerce Catalogues with Self-Refreshing RetrievalFINALLY: A Dataset Recommender for RecSys ExperimentsFusion AutoEncoder: Feature-Grounded Embeddings for Products with Mixed Structured AttributesGroupRec: A Unified Toolkit for Reproducible and Inspectable Group Recommendation ResearchHow Would You Like Your Recommendations? p-book, a Self-Recommending Living Book That Teaches How Recommender Systems WorkLoom: LLM-Powered Naturally Embedded RecommendationLumi: An LLM-Grounded System for Safety-Aware POI RecommendationsPerseus: A Demo of Modular Personalization over Heterogeneous Event SequencesReBait: From Clickbait to Informed Choice in Live News Recommendation FeedsScikit-Rank: Scikit-learn-Compatible Neural Ranking Models for Tabular Recommender SystemsSeqPack: Interactive Composable Sequence Compression for Real-time Recommender SystemsStreamlitRecommenders: Towards Recommendation Inspectability as a New Reproducibility StandardTasteprint: Cross-Platform Recommendation AgentValueVet: Auditing Recommender Systems for Fairness and Value Alignment Using Constitutional AI PrinciplesWiseFood: A Multi-Task Conversational Recommender for Trustworthy, Integrated, and Personalized Dietary GuidanceA Design-Space Approach to Robust Agentic Conversational RecommendationAligning Algorithms with Axiology: Operationalizing Journalistic Values in News Recommender SystemsDiscover Weakly No More: User-Centric Design Criteria for Discovery in Music RecommendationFrom Click Modeling to Offline and Off-Policy Evaluation in Carousel RecommendationHuman Value in Recommendations: Towards Improved Evaluation MeasuresTeaching Recommenders to Listen: Bidirectional Natural-Language Control for Personalized NewsToward Human-Aligned Recommender Systems via Cognitive and Behavioral User ModelingTransparency and Control for Provider-side Recommendation InterfacesUnbiased Recommender Systems with Implicit FeedbackWhat Reasoning Should a Recommender Have? Designing Reasoning for Inference-Time Scaling in RecommendationKeeping the Index Open: The Recommendation-Side Cost of Shared Search and RecommendationReal-Time Hard Negative Sampling via LLM-based Clustering for Large-Scale Two-Tower RetrievalMulti-Modal Semantic Expansion with Constrained LLM Reranking for Conversational Music RecommendationCognitive Biases in Recommender Systems — материалы COBIS RecSys 2026End-to-End User and Item Embeddings: Specializing Retrieval Representations for RankingReward Optimizing Recommender SystemsCognitive Biases in Recommender SystemsCalibrating Reproduced Claims in Recommender SystemsLightweight Ranking Heads: Accelerating Multi-Task Experimentation in Production Recommender SystemsTiming-Aware Repurchase Prediction for Web-Scale E-Commerce: Survival Models for Multi-Surface Grocery Recommendation

Наведите на точку — появится название работы.

05

Как читать эту карту

Покрытие без двойного счёта

В опубликованном списке принятых работ конференции 261 записей. После объединения дублей и вариантов названий остаётся 259 уникальных; ещё 10 материалов нашлись за его пределами — среди воркшоп-докладов, tutorial и отдельных публикаций. В каталоге 269 работ.

У 269 работ есть разбор полного текста, 0 ждут его и в графики не входят. У tutorial полным текстом может быть короткая аннотация — тип материала указан на странице работы. Аннотация tutorial и его proceedings-версия — один материал, а не два независимых доказательства.

Идентичность и темы

Работа опознаётся по DOI, затем по arXiv и подтверждённым вариантам названия; у каждой записи один адрес. Если разбор ссылается на внешнюю статью, это не значит, что она попала в каталог конференции.

Темы — наша разметка разборов, и тем у одной работы может быть несколько. Числа разных датасетов, протоколов и A/B-тестов мы не складываем в один процент улучшения: онлайн-эффект действует в условиях своей статьи. Это обзор того, что мы прочитали, а не обещание покрыть все proceedings.

Zotero: 269 из 269

Ссылки ведут на конкретные записи библиотеки Zotero; запись считается привязанной, только если совпали DOI и название.

Материалы с записью в Zotero →
RecSys2026

Разборы работ RecSys 2026 · обновлено 2026-10-16.