CIKM 2022 |
您所在的位置：网站首页 › deep view swiss › CIKM 2022 |
CIKM 2022
|
作者：孙文奇 @孙文奇，中国人民大学高瓴人工智能学院博士二年级在读，导师为赵鑫教授，研究方向为推荐系统。 引言第31届国际信息与知识管理大会（The 31st ACM International Conference on Information and Knowledge Management, CIKM 2022）计划于2022年10月17日-10月21日以线上线下混合方式召开。ACM CIKM是CCF推荐的B类国际学术会议，是信息检索和数据挖掘领域最重要的学术会议之一。这次会议共录用274篇长文（Full Paper）、91篇应用文（Applied Paper）和196篇短文/资源文（Short / Resource Paper）。官方发布的接收论文列表： 本文选取了CIKM 2022中86篇长文和26篇应用文，重点对推荐系统相关论文（85篇）按不同的任务场景和研究话题进行分类整理，也对其他热门研究方向（预训练模型、信息检索和知识图谱等，27篇）进行了归类，以供参考。文章同步发布在AI Box微信公众号和知乎专栏（微信搜索「RUC AI Box」、知乎搜索「AI Box专栏」），整理过程中难免有疏漏，欢迎大家在文章下方评论留言，交流探讨。 从词云图看今年CIKM的研究热点：根据长文和应用文的标题绘制如下词云图，可以看到今年研究方向主要集中在Recommendation、Retrieval和Knowledge Graph三个方向，也包括Pre-trained Language Model等NLP方向。主要任务包括：Click-Through Rate、Sequential Recommendation、User Modeling等；热门技术包括：Graph Neural Network、Contrastive Learning等，其中基于Sequence和Graph的任务和技术依旧是今年的研究热点。  对于推荐算法的开发与复现，欢迎大家使用推荐系统工具包RecBole（伯乐）。RecBole 是一个基于 PyTorch 实现的，面向研究者的，易于开发与复现的，统一、全面、高效的推荐系统代码库。 工具包： 数据集： 论文（RecBole 2.0已被CIKM 2022录用为Resource Paper）： 本文目录1 按推荐的任务场景划分Click-Through RateCollaborative FilteringSequential/Session-based RecommendationKnowledge-Aware RecommendationSocial RecommendationNews RecommendationText-Aware RecommendationConversational Recommender SystemCross-domain RecommendationOnline RecommendationGroup RecommendationOther Tasks2 按推荐的研究话题划分Debias in Recommender SystemFairness in Recommender SystemExplanation in Recommender SystemCold-start in Recommender SystemRanking in Recommender SystemEvaluationOthers3 热门技术在推荐中的应用Graph Neural Network in Recommender SystemContrastive Learning in Recommender SystemVariational Autoencoder in Recommender SystemMeta/Zero-shot/Few-shot Learning4 其他研究方向Pre-trainingInformation RetrievalKnowledge Graph1. 按推荐的任务场景划分1.1 Click-Through RateTowards Understanding the Overfitting Phenomenon of Deep Click-Through Rate ModelsSparse Attentive Memory Network for Click-through Rate Prediction with Long SequencesGraph Based Long-Term And Short-Term Interest Model for Click-Through Rate PredictionGRP: A Gumbel-based Rating Prediction Framework for Imbalanced RecommendationHierarchically Fusing Long and Short-Term User Interests for Click-Through Rate Prediction in Product SearchOptEmbed: Learning Optimal Embedding Table for Click-through Rate PredictionSampling Is All You Need on Modeling Long-Term User Behaviors for CTR Prediction【applied paper】GIFT: Graph-guIded Feature Transfer for Cold-Start Video Click-Through Rate Prediction【applied paper】1.2 Collaborative FilteringAsymmetrical Context-aware Modulation for Collaborative Filtering RecommendationDynamic Hypergraph Learning for Collaborative FilteringNEST: Simulating Pandemic-like Events for Collaborative Filtering by Modeling User Needs EvolutionMDGCF: Multi-Dependency Graph Collaborative Filtering with Neighborhood- and Homogeneous-level DependenciesITSM-GCN: Informative Training Sample Mining for Graph Convolution Network-based Collaborative Filtering1.3 Sequential/Session-based RecommendationBeyond Learning from Next Item: Sequential Recommendation via Personalized Interest SustainabilityEvolutionary Preference Learning via Graph Nested GRU ODE for Session-based RecommendationDisentangling Past-Future Modeling in Sequential Recommendation via Dual NetworksDual-Task Learning for Multi-Behavior Sequential RecommendationDually Enhanced Propensity Score Estimation in Sequential RecommendationTemporal Contrastive Pre-Training for Sequential RecommendationStorage-saving Transformer for Sequential RecommendationsTime Lag Aware Sequential RecommendationHierarchical Item Inconsistency Signal learning for Sequence Denoising in Sequential RecommendationA Relevant and Diverse Retrieval-enhanced Data Augmentation Framework for Sequential Recommendation【applied paper】1.4 Knowledge-Aware RecommendationImproving Knowledge-aware Recommendation with Multi-level Interactive Contrastive LearningLeveraging Multiple Types of Domain Knowledge for Safe and Effective Drug RecommendationAccurate Action Recommendation for Smart Home via Two-Level Encoders and Commonsense KnowledgeKnowledge Enhanced Multi-Interest Network for the Generation of Recommendation Candidates【applied paper】1.5 Social RecommendationUser Recommendation in Social Metaverse with VR1.6 News RecommendationDeepVT: Deep View-Temporal Interaction Network for News Recommendation1.7 Text-Aware RecommendationReview-Based Domain Disentanglement without Duplicate Users or Contexts for Cross-Domain RecommendationImproving Text-based Similar Product Recommendation for Dynamic Product Advertising at Yahoo【applied paper】1.8 Conversational Recommender SystemRethinking Conversational Recommendations: Is Decision Tree All You Need?Two-level Graph Path Reasoning for Conversational Recommendation with User Realistic Preference1.9 Cross-domain RecommendationContrastive Cross-Domain Sequential RecommendationCross-domain Recommendation via Adversarial AdaptationGromov-Wasserstein Guided Representation Learning for Cross-Domain RecommendationFedCDR: Federated Cross-Domain Recommendation for Privacy-Preserving Rating PredictionCross-Domain Aspect Extraction using Transformers Augmented with Knowledge GraphsAdaptive Domain Interest Network for Multi-domain Recommendation【applied paper】1.10 Online RecommendationKnowledge Extraction and Plugging for Online Recommendation【applied paper】SASNet: Stage-aware sequential matching for online travel recommendation【applied paper】1.11 Group RecommendationGBERT: Pre-training User representations for Ephemeral Group Recommendation1.12 Other TasksMARIO: Modality-Aware Attention and Modality-Preserving Decoders for Multimedia RecommendationTarget Interest Distillation for Multi-Interest RecommendationA Multi-Interest Evolution Story: Applying Psychology in Query-based Recommendation for Inferring Customer IntentionHySAGE: A Hybrid Static and Adaptive Graph Embedding Network for Context-Drifting RecommendationsTask Publication Time Recommendation in Spatial CrowdsourcingAutoMARS: Searching to Compress Multi-Modality Recommendation SystemsMIC: Model-agnostic Integrated Cross-channel Recommender【applied paper】A Case Study in Educational Recommenders: Recommending Music Partitures at Tomplay【applied paper】Real-time Short Video Recommendation on Mobile Devices【applied paper】Scenario-Adaptive and Self-Supervised Model for Multi-Scenario Personalized Recommendation【applied paper】Multi-Faceted Hierarchical Multi-Task Learning for Recommender Systems【applied paper】2. 按推荐的研究话题划分2.1 Debias in Recommender SystemQuantifying and Mitigating Popularity Bias in Conversational Recommender SystemsRepresentation Matters When Learning From Biased Feedback in RecommendationHard Negatives or False Negatives: Correcting Pooling Bias in Training Neural Ranking ModelsUnbiased Learning to Rank with Biased Continuous FeedbackDebiased Balanced Interleaving at Amazon Search【applied paper】Mitigating Biases in Student Performance Prediction via Attention-Based Personalized Federated Learning【applied paper】2.2 Fairness in Recommender SystemRAGUEL: Recourse-Aware Group Unfairness EliminationTowards Principled User-side Recommender Systems2.3 Explanation in Recommender SystemExplanation Guided Contrastive Learning for Sequential Recommendation2.4 Cold-start in Recommender SystemGenerative Adversarial Zero-Shot Learning for Cold-Start News RecommendationAddressing Cold Start in Product Search via Empirical Bayes【applied paper】2.5 Ranking in Recommender SystemRank List Sensitivity of Recommender Systems to Interaction PerturbationsMemory Bank Augmented Long-tail Sequential RecommendationA Biased Sampling Method for Imbalanced Personalized Ranking2.6 EvaluationKuaiRec: A Fully-observed Dataset and Insights for Evaluating Recommender Systems2.7 OthersAn Uncertainty-Aware Imputation Framework for Alleviating the Sparsity Problem in Collaborative FilteringAdapting Triplet Importance of Implicit Feedback for Personalized RecommendationPROPN: Personalized Probabilistic Strategic Parameter Optimization in Recommendations【applied paper】UDM: A Unified Deep Matching Framework in Recommender Systems【applied paper】Approximate Nearest Neighbor Search under Neural Similarity Metric for Large-Scale Recommendation【applied paper】3. 热门技术在推荐中的应用3.1 Graph Neural Network in Recommender SystemSVD-GCN: A Simplified Graph Convolution Paradigm for RecommendationAutomatic Meta-Path Discovery for Effective Graph-Based RecommendationSpatiotemporal-aware Session-based Recommendation with Graph Neural NetworksMulti-Aggregator Time-Warping Heterogeneous Graph Neural Network for Personalized Micro-video RecommendationThe Interaction Graph Auto-encoder Network Based on Topology-aware for Transferable RecommendationPlatoGL: Effective and Scalable Deep Graph Learning System for Graph-enhanced Real-Time Recommendation【applied paper】3.2 Contrastive Learning in Recommender SystemContrastive Learning with Bidirectional Transformers for Sequential RecommendationDomain-Agnostic Constrastive Representations for Learning from Label ProportionsMulti-level Contrastive Learning Framework for Sequential Recommendation3.3 Variational Autoencoder in Recommender SystemContrastVAE: Contrastive Variational AutoEncoder for Sequential Recommendation3.4 Meta/Zero-Shot/Few-Shot LearningTiger: Transferable Interest Graph Embedding for Domain-Level Zero-Shot RecommendationMultimodal Meta-Learning for Cold-Start Sequential Recommendation【applied paper】4. 其他研究方向4.1 Pre-trainingCognize Yourself: Graph Pre-Training via Core Graph Cognizing and DifferentiatingCorpusBrain: Pre-train a Generative Retrieval Model for Knowledge-Intensive Language TasksSemorph: A Morphology Semantic Enhanced Pre-trained Model for Chinese Spam Text DetectionGraph Neural Networks Pretraining Through Inherent Supervision for Molecular Property Prediction【applied paper】Fooling MOSS Detection with Pretrained Language Models【applied paper】4.2 Information RetrievalCROLoss: Towards a Customizable Loss for Retrieval Models in Recommender SystemsContrastive Label Correlation Enhanced Unified Hashing Encoder for Cross-modal RetrievalEvaluating Interpolation and Extrapolation Performance of Neural Retrieval ModelsDetecting Significant Differences Between Information Retrieval Systems via Generalized Linear ModelsPLAID: An Efficient Engine for Late Interaction RetrievalScattered or Connected? An Optimized Parameter-efficient Tuning Approach for Information RetrievalSpaDE: Improving Sparse Representations using a Dual Document Encoder for First-stage RetrievalDense Retrieval with Entity ViewsApproximated Doubly Robust Search Relevance Estimation【applied paper】Cross-Domain Product Search with Knowledge Graph【applied paper】4.3 Knowledge GraphAlong the Time: Timeline-traced Embedding for Temporal Knowledge Graph CompletionExplainable Link Prediction in Knowledge HypergraphsI Know What You Do Not Know: Knowledge Graph Embedding via Co-distillation LearningInductive Knowledge Graph Reasoning for Multi-batch Emerging EntitiesLarge-scale Entity Alignment via Knowledge Graph Merging, Partitioning and EmbeddingContrastive Knowledge Graph Error DetectionContrastive Representation Learning for Conversational Question Answering over Knowledge GraphsDA-Net: Distributed Attention Network for Temporal Knowledge Graph ReasoningDiscovering Fine-Grained Semantics in Knowledge Graph RelationsHigh-quality Task Division for Large-scale Entity AlignmentInteractive Contrastive Learning for Self-Supervised Entity AlignmentCognitive Diagnosis focusing on Knowledge Components【applied paper】
|
【本文地址】