Conversational Recommendation Systems (CRSs) have recently started to leverage pretrained language models (LM) such as BERT for their ability to semantically interpret a wide range of preference statement variations. However, pretrained LMs are prone to intrinsic biases in their training data, which may be exacerbated by biases embedded in domain-specific language data (e.g., user reviews) used to fine-tune LMs for CRSs. We study a simple LM-driven recommendation backbone (termed LMRec) of a CRS to investigate how unintended bias — i.e., bias due to language variations such as name references or indirect indicators of sexual orientation or location that should not affect recommendations — manifests in substantially shifted price and category distributions of restaurant recommendations. For example, offhand mention of names associated with the black community substantially lowers the price distribution of recommended restaurants, while offhand mentions of common male-associated names lead to an increase in recommended alcohol-serving establishments. While these results raise red flags regarding a range of previously undocumented unintended biases that can occur in LM-driven CRSs, there is fortunately a silver lining: we show that train side masking and test side neutralization of non-preferential entities nullifies the observed biases without significantly impacting recommendation performance.

会话推荐系统 (CRS) 最近开始利用 BERT 等预训练语言模型 (LM)，因为它们能够从语义上解释各种偏好陈述的变体。然而，经过预训练的 LM 在其训练数据中容易出现内在偏差，而嵌入在用于为 CRS 微调 LM 的特定领域语言数据（例如，用户评论）中的偏差可能会加剧这种偏差。我们研究了 CRS 的一个简单的 LM 驱动的推荐骨干（称为 LMRec），以调查意外偏见是如何产生的——即，由于语言变化引起的偏见，例如名称引用或性取向或位置的间接指标，这些不应该影响推荐——体现在餐厅推荐的价格和类别分布发生重大变化。例如，随意提及与黑人社区相关的名字会大大降低推荐餐厅的价格分布，而随意提及与男性相关的常见名字会导致推荐的酒类服务机构增加。虽然这些结果对 LM 驱动的 CRS 中可能发生的一系列先前未记录的意外偏差提出了警告，但幸运的是有一线希望：我们表明，训练侧掩蔽和测试侧非优先实体的中和消除了观察到的偏差，而没有显着影响推荐性能。