用反事实方法增强大模型对用户行为序列的因果建模，提升推荐精准度。
Causality-Enhanced Behavior Sequence Modeling in LLMs for Personalized Recommendation
- 通过反事实推理识别行为序列对推荐结果的因果影响。
- 在真实数据集上，推荐准确率显著优于基线方法。
- 适合关注推荐系统可解释性与行为序列建模的研究者。
近期推荐系统研究聚焦于利用大语言模型（LLMs）改进用户偏好建模，取得了良好效果。然而，现有基于LLM的方法未能充分挖掘用户行为序列信息，导致个性化推荐的偏好建模不够理想。本文提出一种新型反事实微调（Counterfactual Fine-Tuning, CFT）方法，通过显式强调行为序列在生成推荐时的作用来解决该问题。具体而言，我们采用反事实推理识别行为序列对模型输出的因果效应，并设计一个基于这些效应直接拟合真实标签的任务，实现显式强调。此外，我们还开发了一种标记级加权机制，调整不同物品标记的强调强度，反映行为序列影响力随时间递减的特性。在多个真实世界数据集上的大量实验表明，CFT能有效提升行为序列建模能力。代码已公开于 https://github.com/itsmeyjt/CFT。
原文摘要 · Abstract (English)
Recent advancements in recommender systems have focused on leveraging Large Language Models (LLMs) to improve user preference modeling, yielding promising outcomes. However, current LLM-based approaches struggle to fully leverage user behavior sequences, resulting in suboptimal preference modeling for personalized recommendations. In this study, we propose a novel Counterfactual Fine-Tuning (CFT) method to address this issue by explicitly emphasizing the role of behavior sequences when generating recommendations. Specifically, we employ counterfactual reasoning to identify the causal effects of behavior sequences on model output and introduce a task that directly fits the ground-truth labels based on these effects, achieving the goal of explicit emphasis. Additionally, we develop a token-level weighting mechanism to adjust the emphasis strength for different item tokens, reflecting the diminishing influence of behavior sequences from earlier to later tokens during predicting an item. Extensive experiments on real-world datasets demonstrate that CFT effectively improves behavior sequence modeling. Our codes are available at https://github.com/itsmeyjt/CFT.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。