Recently, causal inference has attracted increasing attention from researchers of recommender systems (RS), which analyzes the relationship between a cause and its effect and has a wide range of real-world applications in multiple fields. Causal inference can model the causality in recommender systems like confounding effects and deal with counterfactual problems such as offline policy evaluation and data augmentation. Although there are already some valuable surveys on causal recommendations, these surveys introduce approaches in a relatively isolated way and lack theoretical analysis of existing methods. Due to the unfamiliarity with causality to RS researchers, it is both necessary and challenging to comprehensively review the relevant studies from the perspective of causal theory, which might be instructive for the readers to propose new approaches in practice. This survey attempts to provide a systematic review of up-to-date papers in this area from a theoretical standpoint. Firstly, we introduce the fundamental concepts of causal inference as the basis of the following review. Then we propose a new taxonomy from the perspective of causal techniques and further discuss technical details about how existing methods apply causal inference to address specific recommender issues. Finally, we highlight some promising directions for future research in this field.
翻译:最近,因果推断在推荐系统研究中受到越来越多的关注,它分析因果关系并在多个领域具有广泛的应用。因果推断可以对推荐系统中的相关效果进行建模,例如混淆效应,并处理反事实问题,如线下策略评估和数据增强。虽然关于因果推荐已经有一些有价值的综述,但这些综述以相对孤立的方式介绍方法,缺乏现有方法的理论分析。由于推荐系统研究人员对因果关系的陌生,因此从因果理论的角度全面评估相关研究对于提出实践中的新方法既有必要又具有挑战性。本综述试图从理论的角度全面评估此领域内最新的论文。首先,我们介绍因果推断的基本概念,作为以下综述的基础。然后,我们从因果技术的角度提出了一个新的分类法,并进一步讨论了现有方法如何应用因果推断来解决特定的推荐问题的技术细节。最后,我们强调了该领域未来研究的一些有前途的方向。