WWW（World Wide Web Conference，简称 WWW）是中国计算机学会（CCF）推荐的 A 类国际学术会议，与 SIGIR、KDD、WSDM 并列，为 Web 检索、数据挖掘、推荐系统、社交网络、语义 Web、隐私安全等方向最具影响力的发表平台之一。 本文统计了 WWW 2024推荐相关论文共计 64 篇，按不同的主题进行分类，论文内容非常丰富，具体的论文全文信息可参考<www2024.thewebconf.org/accepted/re…
这里收集整理的论文主包含但不限于：推荐/排序/CTR、检索-推荐融合、用户行为建模（序列/多行为/跨域）、多模态、图推荐、可信与公平、联邦/隐私、安全、系统与效率、冷启动与长尾等。
WWW 2024 概览
- 接收情况：投稿 2008 篇，录用 406 篇，录用率 20.2%。
| 序号 | 标题 | 主要思想 |
|---|---|---|
| 1 | Intelligent Model Update Strategy for Sequential Recommendation | 面向序列推荐的模型迭代/更新策略，缓解模型随时间漂移与陈旧问题 |
| 2 | A Data-Centric Multi-Objective Learning Framework for Responsible Recommendation Systems | 数据中心的多目标学习框架，同时优化准确性与责任目标（公平/稳健等） |
| 3 | User Distribution Mapping Modelling with Collaborative Filtering for Cross Domain Recommendation | 用协同过滤建模并对齐跨域用户分布映射，提升跨域推荐迁移效果 |
| 4 | Collaborative Large Language Model for Recommender Systems | 以“协作式”大语言模型增强推荐（多代理/协作推理/协同信号融合） |
| 5 | Rethinking Cross-Domain Sequential Recommendation under Open-World Assumptions | 在开放世界设定下重审跨域序列推荐（未知物品/域变化），提升泛化与鲁棒性 |
| 6 | Temporal Conformity-aware Hawkes Graph Network for Recommendations | 结合 Hawkes 过程与图网络建模时间影响与从众/一致性效应的推荐方法 |
| 7 | Not All Embeddings are Created Equal: Towards Robust Cross-domain Recommendation via Contrastive Learning | 用对比学习做跨域表征对齐与鲁棒增强，缓解跨域 embedding 质量差异 |
| 8 | Harnessing Large Language Models for Text-Rich Sequential Recommendation | 利用LLM 编码文本丰富的行为序列，提升序列推荐的理解与表达能力 |
| 9 | Multi-Modal Knowledge Distillation for Recommendation with Prompt-Tuning | 多模态知识蒸馏 + prompt tuning，把强多模态能力压到轻量推荐模型 |
| 10 | Lower-Left Partial AUC: An Effective and Efficient Optimization Metric for Recommendation | 提出/优化 Lower-Left pAUC 指标，聚焦低误报区域的排序效果与效率，校准优化 |
| 11 | Accurate Cold-start Bundle Recommendation via Popularity-based Coalescence and Curriculum Heating | 面向冷启动组合推荐：用流行度聚合 + 课程式训练提升冷启动准确性 |
| 12 | Efficient Noise-Decoupling for Multi-Behavior Sequential Recommendation | 多行为序列推荐的噪声解耦：分离不同行为的噪声与贡献，提升效率与效果 |
| 13 | Scalable and Provably Fair Exposure Control for Large-Scale Recommender Systems | 大规模推荐的曝光控制：可扩展且具公平性理论保证的分配/调控方法 |
| 14 | Causally Debiased Time-aware Recommendation | 用因果方法进行去偏并结合时间建模，减少时间相关偏差带来的推荐误差 |
| 15 | Uplift Modeling for Target User Attacks on Recommender Systems | 用 uplift modeling 设计/评估对目标用户的攻击增益，研究推荐安全对抗 |
| 16 | Physical Trajectory Inference Attack and Defense in Decentralized POI Recommendation | 针对去中心化 POI 推荐的物理轨迹推断攻击与防御机制 |
| 17 | Graph Contrastive Learning with Kernel Dependence Maximization for Social Recommendation | 社交推荐：图对比学习 + 核依赖最大化，增强社交图表示与推荐信号 |
| 18 | FairSync: Ensuring Amortized Group Exposure in Distributed Recommendation Retrieval | 分布式检索式推荐中的群体曝光公平：以摊还约束方式长期满足公平目标 |
| 19 | Top-Personalized-K Recommendation | 以 Top-K 个性化为目标的推荐建模/优化（强调 top 区域质量或个性化阈值） |
| 20 | Debiasing Recommendation with Personal Popularity | 用“个体流行度”校正偏差，对不同用户的流行度偏好做去偏建模 |
| 21 | UnifiedSSR: A Unified Framework of Sequential Search and Recommendation | 统一建模序列搜索与推荐（SSR），在同一框架下融合 search 与 rec 信号 |
| 22 | Generative News Recommendation | 生成式新闻推荐：用生成模型直接生成/排序候选以提升新闻推荐效果 |
| 23 | MMPOI: A Multi-Modal Content-Aware Framework for POI Recommendations | POI 推荐的多模态内容感知框架，融合多模态信息提升地点推荐 |
| 24 | Representation Learning with Large Language Models for Recommendation | 用 LLM 做推荐表征学习（用户/物品/行为文本化编码），增强语义理解 |
| 25 | Challenging Low Homophily in Social Recommendation | 面向低同质性社交图的推荐挑战与方法改进，提升在弱同质图上的性能 |
| 26 | ReLLa: Retrieval-enhanced Large Language Models for Lifelong Sequential Behavior Comprehension in Recommendation | RAG 增强的 LLM 做长期序列行为理解与持续学习，支撑终身序列推荐 |
| 27 | Learning Category Trees for ID-Based Recommendation: Exploring the Power of Differentiable Vector Quantization | 用可微向量量化学习 ID 推荐的类别树结构，增强离散 ID 表征与层次信息 |
| 28 | Retention Depolarization in Recommender System | 研究推荐导致的留存极化效应并提出缓解策略，兼顾留存与用户多样性 |
| 29 | Linear-Time Graph Neural Networks for Scalable Recommendations | 线性时间复杂度的图神经网络，用于可扩展推荐的高效图建模 |
| 30 | Poisoning Federated Recommender Systems with Fake Users | 联邦推荐的投毒攻击：用虚假用户注入干扰模型训练与推荐结果 |
| 31 | Could Small Language Models Serve as Recommenders? Towards Data-centric Cold-start Recommendation | 探索小语言模型做推荐的可行性，并以数据中心方法提升冷启动推荐 |
| 32 | Disentangling the Long-Term Effects of Recommendations on User Consumption Patterns | 解耦推荐对用户消费模式的长期影响（因果/动态效应），避免短期指标误导 |
| 33 | Online Billion-Scale Recommender Systems with Macro Graph Neural Networks | 面向在线百亿规模推荐：用宏观图神经网络建模大规模结构并实现工程可用 |
| 34 | Prompt-enhanced Federated Content Representation Learning for Cross-domain Recommendation | 跨域联邦推荐：用 prompt 增强内容表征学习，实现隐私约束下的跨域迁移 |
| 35 | Intersectional Two-sided Fairness in Recommendation | 交叉群体 + 双边公平（用户侧/物品侧）的统一建模与约束优化 |
| 36 | Link Recommendation to Augment Influence Diffusion with Provable Guarantees | 用链接推荐增强影响力扩散，并提供可证明的性能/覆盖保证 |
| 37 | When Federated Recommendation Meets Cold-Start Problem: Separating Item Attributes and User Interactions | 联邦冷启动：分离物品属性与用户交互两类信号，提升新物品/新用户场景表现 |
| 38 | Recommender Transformers with Behavior Pathways | Transformer 引入行为通路/分支，对多类型行为做分路建模以提升推荐 |
| 39 | RecDCL: Dual Contrastive Learning for Recommendation | 双重对比学习框架，从多视角/多粒度增强推荐表征与判别能力 |
| 40 | Ensuring User-side Fairness in Dynamic Recommender Systems | 动态推荐中的用户侧公平：在时间维度上控制个体受益/曝光的公平性 |
| 41 | AgentCF: Collaborative Learning with Autonomous Language Agents for Recommender Systems | 用自主语言代理协作学习协同过滤（生成/校验偏好），提升推荐学习与冷启动 |
| 42 | Graph Pretraining and Prompt Learning for Recommendation | 图预训练 + prompt 学习，把图结构知识高效迁移到下游推荐任务 |
| 43 | Mirror Gradient: Towards Robust Multimodal Recommender Systems via Exploring Flat Local Minima | 通过镜像梯度探索平坦极小值，提高多模态推荐的鲁棒性与稳定性 |
| 44 | Modeling Balanced Explicit and Implicit Relations with Contrastive Learning for Knowledge Concept Recommendation in MOOCs | MOOC 知识点推荐：用对比学习平衡显式/隐式关系建模，提升概念推荐质量 |
| 45 | Learning Counterfactual Explanations for Recommender Systems | 学习生成反事实解释：告诉用户“改什么行为/特征会改变推荐结果” |
| 46 | Category-based and Popularity-guided Video Game Recommendation: A Balance-oriented Framework | 游戏推荐：以类别与流行度引导实现平衡（兼顾命中率与多样性/新颖性） |
| 47 | Unleashing the Power of Knowledge Graph for Recommendation via Invariant Learning | 知识图推荐：用不变学习提取稳定因果/语义信号，提升跨场景泛化 |
| 48 | Enhancing Recommendation Accuracy and Diversity with Box Embedding: A Universal Framework | 用 Box Embedding 统一刻画包含关系与不确定性，改善准确性-多样性权衡 |
| 49 | Leave No One Behind: Online Self-Supervised Self-distillation for Sequential Recommendation | 在线自监督自蒸馏的序列推荐，提升长尾/弱监督样本下的学习与覆盖 |
| 50 | Distributionally Robust Graph-based Recommendation System | 分布鲁棒优化（DRO）用于图推荐，提升在分布偏移/噪声下的稳健性 |
| 51 | Doubly Calibrated Estimator for Recommendation on Data Missing Not At Random | 面向 MNAR 缺失机制的双重校准估计器，提升推荐评估/学习的无偏性 |
| 52 | Reconciling the accuracy-diversity trade-off in recommendations | 系统性缓解准确性-多样性矛盾，通过目标/约束/重排实现折中最优 |
| 53 | Co-clustering for Federated Recommender System | 联邦推荐中的协同聚类（用户/物品共聚类）以提升个性化与隐私友好训练 |
| 54 | M-scan: A Multi-Scenario Causal-driven Adaptive Network for Recommendation | 多场景因果驱动自适应网络，解决场景差异与迁移偏差，提升多场景推荐 |
| 55 | Is Contrastive Learning Necessary? A Study of Data Augmentation vs Contrastive Learning in Sequential Recommendation | 序列推荐实证研究：对比数据增强与对比学习的贡献，回答“是否必须对比学习” |
| 56 | Can Small Language Models be Good Reasoners in Recommender Systems? | 探索小语言模型在推荐任务中的推理能力与适用边界（何时能当“会想的模型”） |
| 57 | Negative Sampling in Next-POI Recommendations: Observation, Approach, and Evaluation | Next-POI 推荐的负采样系统分析：提出改进策略并给出评测方法 |
| 58 | Towards Personalized Privacy: User-Governed Data Contribution for Federated Recommendation | 联邦推荐的个性化隐私：用户可控的数据贡献机制，实现隐私-效果可调 |
| 59 | Federated Heterogeneous Graph Neural Network for Privacy-preserving Recommendation | 隐私保护推荐：联邦异构图神经网络建模多类型关系，在不共享原始数据下训练 |
| 60 | Decentralized Collaborative Learning with Adaptive Reference Data for On-Device POI Recommendation | 端侧 POI 推荐的去中心化协同学习，引入自适应参考数据提升本地训练 |
| 61 | Towards Efficient Communication and Secure Federated Recommendation System via Low-rank Training | 用低秩训练降低联邦推荐通信开销并增强安全性/稳健性，兼顾效率与防护 |
| 62 | Hierarchical Graph Signal Processing for Collaborative Filtering | 将层次化图信号处理引入协同过滤，在频域/多尺度上提取协同信号 |
| 63 | General Debiasing for Graph-based Collaborative Filtering via Adversarial Graph Dropout | 通过对抗式图 dropout 做通用去偏，减轻图协同过滤中的结构/采样偏差 |
