CIKM 2021 | 推荐系统相关论文分类整理

客官°小女子只卖身不卖艺 2022-08-28 04:46 209阅读 0赞

机构｜中国人民大学高瓴人工智能学院

研究方向｜推荐系统

本文选取了CIKM2021中85篇长文、15篇应用文和29篇短文，**重点对推荐系统相关论文（76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、知识图谱等，53篇）进行了归类**。**文章也同步发布在****AI****Box****知乎专栏（知乎搜索 AI Box专栏），欢迎大家在知乎专栏的文章下方评论留言，交流探讨！**

# 引言 #

第30届国际信息与知识管理大会（The 30th ACM International Conference on Information and Knowledge Management, CIKM 2021）计划于2021年11月1日-11月5日在线召开。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。这次会议共收到1251篇长文（Full paper）、290篇应用文（Applied paper）和626篇短文（Short paper）投稿，有271篇长文、69篇应用文和178篇短文被录用，录用率分别为21.7%、23.8%和28.4%。官方发布的接收论文列表：

http://www.cikm2021.org/accepted-papers

本文选取了CIKM 2021中85篇长文、15篇应用文和29篇短文，**重点对推荐系统相关论文（76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、知识图谱等，53篇）进行了归类，**以供参考。

从词云图看**今年CIKM的研究热点：**根据长文、应用文和短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、Conversation等NLP方向。主要任务包括：Click-Through Rate、Sequential Recommendation、Knowledge Graph Embedding、User Modeling等；热门技术包括：Graph Neural Network、Contrastive Learning、Transformer、Attention等，其中基于Graph的任务和技术依旧是今年的研究热点。

![29456f88ab128aba49144516585198e9.png][]

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。RecBole目前已经有72个常见的推荐模型和28个常用的推荐系统数据集。

*  工具包：
    
    https://github.com/RUCAIBox/RecBole
 *  数据集：
    
    https://github.com/RUCAIBox/RecSysDatasets
 *  论文（RecBole已被CIKM 2021录用为Resource paper）：  
    RecBole: Towards a Unified, Comprehensive and Efficient Framework for Recommendation Algorithms
    
    ![8bb97e6534de4a7f3cbe77b33529a922.png][]

# 本文目录 #

## 1 按推荐的任务场景划分 ##

*  Click-Through Rate
 *  Collaborative Filtering
 *  Sequential/Session-based Recommendation
 *  Knowledge-Aware Recommendation
 *  Social Recommendation
 *  News Recommendation
 *  Text-Aware Recommendation
 *  Conversational Recommender System
 *  Cross-domain Recommendation
 *  Point-of-Interest
 *  Online Recommendation
 *  Group Recommendation
 *  Other Tasks

## 2 按推荐的研究话题划分 ##

*  Debias in Recommender System
 *  Fairness in Recommender System
 *  Explanation in Recommender System
 *  Cold-start in Recommender System
 *  Ranking in Recommender System
 *  Evaluation
 *  Others

## 3 热门技术在推荐中的应用 ##

*  Graph Neural Network in Recommender System
 *  Contrastive Learning in Recommender System
 *  Reinforcement Learning in Recommender System
 *  Variational Autoencoder in Recommender System
 *  Zero-Shot Learning in Recommender System

## 4 其他研究方向 ##

*  Pre-training
 *  Transformer
 *  Knowledge Graph
 *  Multi-Modality
 *  Data Augmentation
 *  Meta Learning
 *  Few-Shot Learning

# 1. 按推荐的任务场景划分 #

## 1.1 Click-Through Rate ##

1.  Multi-task Learning for Bias-Free Joint CTR Prediction and Market Price Modeling in Online Advertising【在线广告无偏差联合CTR预估和市场价格建模的多任务学习】
2.  Enhancing Explicit and Implicit Feature Interactions via Information Sharing for Parallel Deep CTR Models【applied paper，用于并行 CTR 的显式和隐式特征交互增强】
3.  TSI: An Ad Text Strength Indicator using Text-to-CTR and Semantic-Ad-Similarity【applied paper，使用 Text-to-CTR 和 Semantic-Ad-Similarity 的广告文本强度指标】
4.  One Model to Serve All: Star Topology Adaptive Recommender for Multi-Domain CTR Prediction【applied paper，用于多领域CTR预估的自适应推荐】
5.  Efficient Learning to Learn a Robust CTR Model for Web-scale Online Sponsored Search Advertising【applied paper，用于在线搜索广告的CTR模型】
6.  AutoIAS: Automatic Integrated Architecture Searcher for Click-Trough Rate Prediction【CTR预估的自动集成搜索架构】
7.  Click-Through Rate Prediction with Multi-Modal Hypergraphs【使用多模态超图的点击率预测】
8.  Open Benchmarking for Click-Through Rate Prediction【开源CTR预估Benchmark】
9.  Disentangled Self-Attentive Neural Networks for Click-Through Rate Prediction【short paper，用于CTR预估的自注意力网络】
10. AutoHERI: Automated Hierarchical Representation Integration for Post-Click Conversion Rate Estimation【short paper，用于点击后转换率估计的分层表示学习】

## 1.2 Collaborative Filtering ##

1.  SimpleX: A Simple and Strong Baseline for Collaborative Filtering【将Cosine Contrastive Loss引入协同过滤】
2.  Incremental Graph Convolutional Network for Collaborative Filtering【增量图卷积神经网络用于协同过滤】
3.  LT-OCF: Learnable-Time ODE-based Collaborative Filtering【Learnable-Time CF】
4.  CausCF: Causal Collaborative Filtering for Recommendation Effect Estimation【applied paper，因果关系协同过滤用于推荐效果评估】
5.  Vector-Quantized Autoencoder With Copula for Collaborative Filtering【short paper，用于协同过滤的矢量量化自动编码器】
6.  Anchor-based Collaborative Filtering for Recommender Systems【short paper，Anchor-based推荐系统协同过滤】

## 1.3 Sequential/Session-based Recommendation ##

1.  Seq2Bubbles: Region-Based Embedding Learning for User Behaviors in Sequential Recommenders【序列推荐中基于区域的用户行为Embedding学习】
2.  Enhancing User Interest Modeling with Knowledge-Enriched Itemsets for Sequential Recommendation【序列推荐中使用物品集增强用户兴趣建模】
3.  Continuous-Time Sequential Recommendation with Temporal Graph Collaborative Transformer【将时序图协同Transformer用于连续时间序列推荐】
4.  Extracting Attentive Social Temporal Excitation for Sequential Recommendation【提取时序激励用于序列推荐】
5.  Hyperbolic Hypergraphs for Sequential Recommendation【使用双曲超图进行序列推荐】
6.  Learning Dual Dynamic Representations on Time-Sliced User-Item Interaction Graphs for Sequential Recommendation【用于序列推荐的在时间片用户物品交互图上的对偶动态表示】
7.  Lightweight Self-Attentive Sequential Recommendation【使用CNN捕获局部特征，使用Self-Attention捕获全局特征】
8.  What is Next when Sequential Prediction Meets Implicitly Hard Interaction?【序列预测与交互】
9.  Modeling Sequences as Distributions with Uncertainty for Sequential Recommendation【short paper，序列建模】
10. Locker: Locally Constrained Self-Attentive Sequential Recommendation【short paper，局部约束的自注意力序列推荐】
11. CBML: A Cluster-based Meta-learning Model for Session-based Recommendation【用于会话推荐的基于聚类的元学习】
12. Self-Supervised Graph Co-Training for Session-based Recommendation【用于会话推荐的自监督图协同训练】

## 1.4 Knowledge-Aware Recommendation ##

1.  A Knowledge-Aware Recommender with Attention-Enhanced Dynamic Convolutional Network【动态卷积用于知识感知的推荐】
2.  Entity-aware Collaborative Relation Network with Knowledge Graph for Recommendation【short paper，KG+RS】
3.  Conditional Graph Attention Networks for Distilling and Refining Knowledge Graphs in Recommendation【GNN+KG+RS】

## 1.5 Social Recommendation ##

1.  Social Recommendation with Self-Supervised Metagraph Informax Network【使用自监督元图网络的社交推荐】

## 1.6 News Recommendation ##

1.  WG4Rec: Modeling Textual Content with Word Graph for News Recommendation【使用Word Graph为新闻推荐建模文本内容】
2.  Popularity-Enhanced News Recommendation with Multi-View Interest Representation【多视角兴趣学习的流行度增强的新闻推荐】
3.  Prioritizing Original News on Facebook【applied paper，原创新闻优先级排序】

## 1.7 Text-Aware Recommendation ##

1.  Counterfactual Review-based Recommendation【基于评论的反事实推荐】
2.  Review-Aware Neural Recommendation with Cross-Modality Mutual Attention【short paper，文本+RS+跨模态】

## 1.8 Conversational Recommender System ##

1.  Popcorn: Human-in-the-loop Popularity Debiasing in Conversational Recommender Systems【采用人在回路方式进行对话推荐系统的流行度去偏】
2.  A Neural Conversation Generation Model via Equivalent Shared Memory Investigation【对话生成】

## 1.9 Cross-domain Recommendation ##

1.  Expanding Relationship for Cross Domain Recommendation【扩展跨领域推荐的关系】
2.  Learning Representations of Inactive Users: A Cross Domain Approach with Graph Neural Networks【short paper，跨领域方法结合图神经网络用于学习非活跃用户表示】
3.  Low-dimensional Alignment for Cross-Domain Recommendation【short paper，跨领域推荐的低维对齐】

## 1.10 Point-of-Interest ##

1.  Answering POI-recommendation Questions using Tourism Reviews【使用旅游者评论回答POI问题】
2.  SNPR: A Serendipity-Oriented Next POI Recommendation Model【面向偶然性的POI推荐】
3.  ST-PIL: Spatial-Temporal Periodic Interest Learning for Next Point-of-Interest Recommendation【short paper，用于POI推荐的时空周期兴趣学习】

## 1.11 Online Recommendation ##

1.  Generative Inverse Deep Reinforcement Learning for Online Recommendation【用于在线推荐的生成式逆强化学习】

## 1.12 Group Recommendation ##

1.  Double-Scale Self-Supervised Hypergraph Learning for Group Recommendation【用于群组推荐的自监督超图学习】
2.  DeepGroup: Group Recommendation with Implicit Feedback【short paper，隐式反馈的群组推荐】

## 1.13 Other Tasks ##

1.  Learning An End-to-End Structure for Retrieval in Large-Scale Recommendations【在大规模推荐中学习一个端到端的结构用于检索】
2.  USER: A Unified Information Search and Recommendation Model based on Integrated Behavior Sequence【基于集成行为序列的统一搜索与推荐模型】
3.  Cross-Market Product Recommendation【跨市场产品推荐】
4.  Multi-hop Reading on Memory Neural Network with Selective Coverage for Medication Recommendation【药物推荐】
5.  Concept-Aware Denoising Graph Neural Network for Micro-Video Recommendation【用于微视频推荐的去噪GNN】

# 2. 按推荐的研究话题划分 #

## 2.1 Debias in Recommender System ##

1.  CauSeR: Causal Session-based Recommendations for Handling Popularity Bias【short paper，用于流行度去偏的因果关系序列推荐】
2.  Mixture-Based Correction for Position and Trust Bias in Counterfactual Learning to Rank【位置和信任偏差】
3.  Unbiased Filtering of Accidental Clicks in Verizon Media Native Advertising【applied paper，广告意外点击的无偏过滤】

## 2.2 Fairness in Recommender System ##

1.  SAR-Net: A Scenario-Aware Ranking Network for Personalized Fair Recommendation in Hundreds of Travel Scenarios【applied paper，用于个性化公平推荐的场景感知排名网络】

## 2.3 Explanation in Recommender System ##

1.  Counterfactual Explainable Recommendation【反事实可解释推荐】
2.  On the Diversity and Explainability of Recommender Systems: A Practical Framework for Enterprise App Recommendation【applied paper，推荐系统的多样性和可解释性】
3.  You Are What and Where You Are: Graph Enhanced Attention Network for Explainable POI Recommendation【applied paper，Attention图神经网络用于可解释推荐】
4.  XPL-CF: Explainable Embeddings for Feature-based Collaborative Filtering【short paper，可解释Embedding用于基于特征的协同过滤】
5.  Grad-SAM: Explaining Transformers via Gradient Self-Attention Maps【short paper，通过梯度Self-Attention解释Transformer】

## 2.4 Cold-start in Recommender System ##

1.  CMML: Contextual Modulation Meta Learning for Cold-Start Recommendation【元学习+冷启动】
2.  Reinforcement Learning to Optimize Lifetime Value in Cold-Start Recommendation【增强学习+冷启动】

## 2.5 Ranking in Recommender System ##

1.  Top-N Recommendation with Counterfactual User Preference Simulation【反事实用户偏好模拟的Top-N推荐】

## 2.6 Evaluation ##

1.  Evaluating Human-AI Hybrid Conversational Systems with Chatbot Message Suggestions【人机混合对话系统评估】
2.  POSSCORE: A Simple Yet Effective Evaluation of Conversational Search with Part of Speech Labelling【使用部分语音标签对会话搜索进行简单有效的评估】

## 2.7 Others ##

1.  DSKReG: Differentiable Sampling on Knowledge Graph for Recommendation with Relational GNN【short paper，用于推荐的知识图谱采样】
2.  Disentangling Preference Representations for Recommendation Critiquing with ?-VAE【用于推荐的VAE偏好表示】

# 3. 热门技术在推荐中的应用 #

## 3.1 Graph Neural Network in Recommender System ##

1.  UltraGCN: Ultra Simplification of Graph Convolutional Networks for Recommendation【GNN+RS】
2.  How Powerful is Graph Convolution for Recommendation?【GNN+RS】

## 3.2 Contrastive Learning in Recommender System ##

1.  Contrastive Curriculum Learning for Sequential User Behavior Modeling via Data Augmentation【applied paper，通过数据增强进行序列用户行为建模的对比课程学习】
2.  Graph Structure Aware Contrastive Knowledge Distillation for Incremental Learning in Recommender Systems【short paper，推荐系统中用于增量学习的图结构感知的对比知识蒸馏】

## 3.3 Reinforcement Learning in Recommender System ##

1.  Explore, Filter and Distill: Distilled Reinforcement Learning in Recommendation【applied paper，推荐中的蒸馏强化学习】

## 3.4 Variational Autoencoder in Recommender System ##

1.  Semi-deterministic and Contrastive Variational Graph Autoencoder for Recommendation【用于推荐的半确定性和对比变分图自动编码器】

## 3.5 Zero-Shot Learning in Recommender System ##

1.  Zero Shot on the Cold-Start Problem: Model-Agnostic Interest Learning for Recommender Systems【零样本学习+冷启动】

# 4. 其他研究方向 #

## 4.1 Pre-training ##

1.  Pre-training for Ad-hoc Retrieval: Hyperlink is Also You Need【Ad-hoc检索预训练】
2.  Pulling Up by the Causal Bootstraps: Causal Data Augmentation for Pre-training Debiasing【用于预训练去偏的因果关系数据增强】
3.  Contrastive Pre-Training of GNNs on Heterogeneous Graphs【图神经网络的对比预训练】
4.  HORNET: Enriching Pre-trained Language Representations with Heterogeneous Knowledge Sources【异构知识来源的预训练】
5.  WebKE: Knowledge Extraction from Semi-structured Web with Pre-trained Markup Language Model【知识抽取+预训练】
6.  Natural Language Understanding with Privacy-Preserving BERT【NLU+BERT】
7.  K-AID: Enhancing Pre-trained Language Models with Domain Knowledge for Question Answering【applied paper，QA+领域知识+预训练】
8.  DialogueBERT: A Self-Supervised Learning based Dialogue Pre-training Encoder【short paper，自监督对话预训练】
9.  BERT-QPP: Contextualized Pre-trained transformers for Query Performance Prediction【short paper，用于查询性能预测的上下文预训练】
10. CANCN-BERT: A Joint Pre-Trained Language Model for Classical and Modern Chinese【short paper，古典和现代中文的联合预训练】
11. Distilling Knowledge from BERT into Simple Fully Connected Neural Networks for Efficient Vertical Retrieval【applied paper，知识蒸馏+预训练+检索】
12. Adversarial Reprogramming of Pretrained Neural Networks for Fraud Detection【short paper，用于欺诈检测的预训练对抗再编程】
13. Adversarial Domain Adaptation for Cross-lingual Information Retrieval with Multilingual BERT【short paper，使用多语言 BERT 进行跨语言信息检索的对抗域自适应】
14. Multi-modal Dictionary BERT for Cross-modal Video Search in Baidu Advertising【applied paper，百度广告中用于跨模态视频搜索的多模态词典BERT】
15. RABERT: Relation-Aware BERT for Target-Oriented Opinion Words Extraction【short paper，用于词提取的关系感知BERT】

## 4.2 Transformer ##

1.  LiteGT: Efficient and Lightweight Graph Transformers【高效轻量化图Transformer】
2.  Block Access Pattern Discovery via Compressed Full Tensor Transformer【Transformer压缩】
3.  Mixed Attention Transformer for Leveraging Word-Level Knowledge to Neural Cross-Lingual Information Retrieval【用于跨语言信息检索的混合注意力Transformer】
4.  Match-Ignition: Plugging PageRank into Transformer for Long-form Text Matching【PageRank+Transformer】
5.  DCAP: Deep Cross Attentional Product Network for User Response Prediction【用于用户响应预测的交叉注意力产品网络】

## 4.3 Knowledge Graph ##

1.  Tracking Semantic Evolutionary Changes in Large-Scale Ontological Knowledge Bases【大规模本体知识库中语义演化的跟踪】
2.  Cycle or Minkowski: Which is More Appropriate for Knowledge Gragh Embedding?【KG Embedding】
3.  HopfE: Knowledge Graph Representation Learning using Inverse Hopf Fibrations【知识图谱表示学习】
4.  Automated Query Graph Generation for Querying Knowledge Graphs【用于查询知识图谱的自动查询图生成】
5.  Differentially Private Federated Knowledge Graphs Embedding【差异化隐私联邦KG Embedding】
6.  A Lightweight Knowledge Graph Embedding Framework for Efficient Inference and Storage【轻量化KG Embedding】
7.  Predicting Instance Type Assertions in Knowledge Graphs Using Stochastic Neural Networks【知识图谱中的实例类型断言预测】
8.  When Hardness Makes a Difference: Multi-Hop Knowledge Graph Reasoning over Few-Shot Relations【小样本关系上的知识图谱多跳推理】
9.  Query Reformulation for Descriptive Queries of Jargon Words Using a Knowledge Graph based on a Dictionary【使用基于字典的知识图谱进行查询重构】
10. Computing and Maintaining Provenance of Query Result Probabilities in Uncertain Knowledge Graphs【不确定知识图谱中计算和维护查询结果概率】
11. REFORM: Error-Aware Few-Shot Knowledge Graph Completion【错误感知的小样本知识图谱补全】
12. DisenKGAT: Knowledge Graph Embedding with Disentangled Graph Attention Network【KG Embedding+GNN】
13. Complex Temporal Question Answering on Knowledge Graphs【QA+KG】
14. Mixed Attention Transformer for Leveraging Word-Level Knowledge to Neural Cross-Lingual Information Retrieval【Transformer+IR】
15. Knowledge Graph Representation Learning as Groupoid: Unifying TransE, RotatE, QuatE, ComplEx【知识图谱表示学习】
16. DataType-Aware Knowledge Graph Representation Learning in Hyperbolic Space【双曲空间中基于数据类型的知识图谱表示学习】
17. Evidential Relational-Graph Convolutional Networks for Entity Classification in Knowledge Graphs【short paper，GNN+KG】

## 4.4 Multi-Modality ##

1.  Student Can Also be a Good Teacher: Extracting Knowledge from Vision-and-Language Model for Cross-Modal Retrieval【short paper，用于跨模态检索的知识提取】
2.  Supervised Contrastive Learning for Multimodal Unreliable News Detection in COVID-19 Pandemic【short paper，用于多模态不可靠新闻检测的有监督对比学习】

## 4.5 Data Augmentation ##

1.  Influence-guided Data Augmentation for Neural Tensor Completion【用于张量补全的数据增强】
2.  Learning to Augment Imbalanced Data for Re-ranking Models【用于再排序模型的数据增强】
3.  Action Sequence Augmentation for Early Graph-based Anomaly Detection【用于异常检测的动作序列增强】

## 4.6 Meta Learning ##

1.  Multimodal Graph Meta Contrastive Learning【short paper，多模态元图对比学习】
2.  Meta-Learning Based Hyper-Relation Feature Modeling for Out-of-Knowledge-Base Embedding【基于元学习的超关系特征建模】
3.  HetMAML: Task-Heterogeneous Model-Agnostic Meta-Learning for Few-Shot Learning Across Modalities【Meta Learning+Few-Shot Learning】
4.  Pruning Meta-Trained Networks for On-Device Adaptation【用于设备自适应的元训练网络剪枝】
5.  Meta Hyperparameter Optimization with Adversarial Proxy Subsets Sampling【元超参优化】

## 4.7 Few-Shot Learning ##

1.  Behind the Scenes: An Exploration of Trigger Biases Problem in Few-Shot Event Classification【小样本学习中偏差问题的探讨】
2.  Learning Discriminative and Unbiased Representations for Few-Shot Relation Extraction【用于小样本关系提取的无偏表示学习】
3.  Multi-view Interaction Learning for Few-Shot Relation Classification【用于小样本关系分类的多视角交互学习】
4.  One-shot Transfer Learning for Population Mapping【单样本迁移学习】
5.  Boosting Few-shot Abstractive Summarization with Auxiliary Tasks【short paper，使用辅助任务提升小样本摘要】
6.  Multi-objective Few-shot Learning for Fair Classification【short paper，Few-Shot Learning+Classification】

[29456f88ab128aba49144516585198e9.png]: /images/20220828/48a7d461df224628b18d9426a0b6f785.png
[8bb97e6534de4a7f3cbe77b33529a922.png]: /images/20220828/5cfe2d7e1ae7441f8555e8819790b160.png