IJCAI2021中知识图谱与推荐系统相关论文如下：知识图谱相关论文Neighborhood Intervention Consistency: Measuring Confidence for Knowledge Graph Link Prediction提出邻域干预一致性方法，用于衡量知识图谱链接预测的置信度，通过分析邻域结构改进预测可靠性。HIP Network: Historical Information Passing Network for Extrapolation Reasoning on Temporal Knowledge Graph设计历史信息传递网络（HIP Network），解决时序知识图谱上的外推推理问题，捕捉时间动态性以支持未来事件预测。Keyword-Based Knowledge Graph Exploration Based on Quadratic Group Steiner Trees基于二次组斯坦纳树算法，实现关键词驱动的知识图谱探索，优化多关键词路径查询效率。Learning Embeddings from Knowledge Graphs With Numeric Edge Attributes针对带数值型边属性的知识图谱，提出嵌入学习方法，保留结构与数值信息的联合表示。Unsupervised Knowledge Graph Alignment by Probabilistic Reasoning and Semantic Embedding结合概率推理与语义嵌入，实现无监督知识图谱对齐，解决跨图谱实体匹配问题。推荐系统相关论文Self-Supervised Adversarial Distribution Regularization for Medication Recommendation引入自监督对抗分布正则化方法，优化药物推荐模型，提升推荐鲁棒性与个性化水平。AMEIR: Automatic Behavior Modeling, Interaction Exploration and MLP Investigation in the Recommender System提出AMEIR框架，集成自动行为建模、交互探索与多层感知机（MLP）分析，增强推荐系统可解释性。Discovering Collaborative Signals for Next POI Recommendation with Iterative Seq2Graph Augmentation通过迭代Seq2Graph增强技术，挖掘兴趣点（POI）推荐中的协作信号，提升序列推荐准确性。MFNP: A Meta-optimized Model for Few-shot Next POI Recommendation构建元优化模型MFNP，解决少样本场景下的兴趣点推荐问题，适应数据稀缺环境。Does Every Data Instance Matter? Enhancing Sequential Recommendation by Eliminating Unreliable Data分析数据实例可靠性，通过剔除不可信数据优化序列推荐模型性能。SafeDrug: Dual Molecular Graph Encoders for Recommending Effective and Safe Drug Combinations提出双分子图编码器（SafeDrug），同时优化药物组合的有效性与安全性，应用于医疗推荐。DA-GCN: A Domain-aware Attentive Graph Convolution Network for Shared-account Cross-domain Sequential Recommendation设计领域感知注意力图卷积网络（DA-GCN），解决共享账户跨域序列推荐中的领域适配问题。UNBERT: User-News Matching BERT for News Recommendation基于BERT模型构建用户-新闻匹配框架（UNBERT），提升新闻推荐的语义理解能力。Pattern-enhanced Contrastive Policy Learning Network for Sequential Recommendation引入模式增强对比策略学习网络，优化序列推荐中的用户行为模式捕捉。Improving Sequential Recommendation Consistency with Self-Supervised Imitation通过自监督模仿学习，提高序列推荐的一致性，减少用户历史行为与推荐结果的偏差。Exploring Periodicity and Interactivity in Multi-Interest Framework for Sequential Recommendation在多兴趣序列推荐框架中探索周期性与交互性，提升长期依赖建模能力。User-as-Graph: User Modeling with Heterogeneous Graph Pooling for News Recommendation提出异质图池化用户建模方法（User-as-Graph），增强新闻推荐中的用户特征表示。Preference-Adaptive Meta-Learning for Cold-Start Recommendation开发偏好适应性元学习框架，解决推荐系统冷启动问题，快速适配新用户或物品。