MERIT：缓解生成式极端多标签分类（XMC）中用户兴趣倾向建模的曝光偏差
MERIT: Mitigating Exposure Bias in Generative XMC for User-Interest Propensity Modeling
查看机构详情
- Carnegie Mellon University（卡内基梅隆大学）
- Amazon（亚马逊公司）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
针对电商用户兴趣匹配的曝光偏差问题，提出MERIT框架，通过自校正目标缓解偏差，在25万+兴趣类数据集上提升召回率等指标，生产中用户转化率提升0.26%。
中文摘要 AI 辅助
大规模将用户与兴趣类别匹配是个性化购物的核心，然而在大型电商平台中该任务极具挑战性，因为标签空间不断演变，且用户兴趣信号稀疏且呈长尾分布。自回归语言模型颇具吸引力，因为其世界知识和描述符语义先验可在极端标签空间中泛化，并支持多个有效标签分配。但在教师强制微调下，推理时的预测会成为条件上下文的一部分：早期错误会引导后续输出向共现标签偏移，过度生成近邻相关标签，且遗漏不相关的真实兴趣。我们提出MERIT，这是一种用户兴趣倾向建模框架，通过自校正目标缓解曝光偏差。针对黄金标签与挖掘的难负标签的打乱混合样本，采用置换不变的多目标损失，使生成器接触错误前缀，同时保留教师强制训练的效率。该训练目标将监督集中在分类位置，产生与倾向对齐的隐藏状态，为轻量评分器提供支持，用于双向检索（用户对应兴趣、兴趣对应用户）。在拥有25万+兴趣类别的专有电商数据集上，MERIT将全局召回率至少提升11.9%，平均Hit@k提升6.1%；在生产环境A/B测试中，用户转化率提升0.26%。
英文摘要
Matching users to interest categories at scale is central to personalized shopping, but the task is challenging in large e-commerce platforms, where label spaces continually evolve and user-interest signals are sparse and long-tailed. Autoregressive language models are appealing because their world knowledge and semantic priors over descriptors generalize across extreme label spaces and accommodate multiple valid label assignments. Yet under teacher-forced fine-tuning, inference-time predictions become part of the conditioning context: early errors steer later outputs toward co-occurring labels, over-generating near-correlates and missing unrelated true interests. We present MERIT, a framework for user-interest propensity modeling that mitigates this exposure bias through a self-correction objective. A permutation-invariant multi-target loss over shuffled mixtures of gold and mined hard-negative labels exposes the generator to erroneous prefixes while preserving the efficiency of teacher-forced training. This training objective concentrates supervision at classification positions, yielding propensity-aligned hidden states powering a lightweight scorer for bidirectional retrieval (interests for users and users for interests). On a proprietary e-commerce dataset with 250k+ interest categories, MERIT improves global recall by at least 11.9% and average Hit@k by 6.1%. In production A/B tests, it achieves +0.26% gain in user conversion.
