Collected data with non-random missing labels poses a widely recognized challenge for unbiased learning. For example, in recommender systems, users are free to choose whether or not to rate an item. To achieve unbiased learning under MNAR data, a variety of methods have been proposed, such as reweighting and imputation. Among them, doubly robust (DR) based methods are widely adopted due to their appealing theoretical guarantees. However, these guarantees rely on strong assumptions that either the propensity or the imputation is accurate for all units (such as user-item pairs), which is very hard to achieve in real-world scenarios. Previous studies show that a small error in imputation can lead to a large bias in DR-based methods. Furthermore, for units with missing labels, we lack an effective method to evaluate the imputation quality. In this work, we propose a model-agnostic framework to assess the accuracy of imputed labels and to correct imputations with large bias based on conformal prediction. Specifically, we leverage conformal prediction to construct a valid prediction set for units with unobserved labels, and revise imputations that fall outside this set. Extensive experiments are conducted on three real-world datasets and one semi-synthetic dataset to show the effectiveness of our proposed method. Our code is available at https://github.com/lixiang-222/conformal-prediction-for-MNAR.
Chunyuan Zheng, Xiang Li, Hang Pan et al.· Proceedings of the 32nd ACM...· 0 citations
Large language models (LLMs) have recently emerged as powerful backbones for recommender systems by reformulating recommendation as a token-level generation task. Despite their promise, we identify a pervasive yet underexplored issue: $\textit{Length Bias}$. Because items are represented by textual descriptions of varying lengths, LLM-based recommenders can be systematically biased in two ways. On the input side, longer item descriptions occupy more tokens in the context and thus receive disproportionately large aggregate attention mass during user preference modeling. On the output side, decoding based on summed autoregressive log-likelihood score inherently disfavors long items. Worse still, conventional length normalization can introduce an additional bias and even degrade recommendation performance. To address this problem, we propose $\textbf{LBR}$ ($\textbf{L}$ength $\textbf{B}$ias $\textbf{R}$eduction), a lightweight and model-agnostic framework for mitigating length bias in LLM-based recommendation. LBR mitigates input-side bias via Length-Aware Attention Calibration, which incorporates a length-dependent offset into attention logits to neutralize attention skew. For the output side, LBR introduces Effective Information Length Normalization, replacing naive token count with an information-theoretic length surrogate derived from the branching structure of the prefix tree. Extensive experiments on three real-world Amazon datasets and two representative LLM-based recommenders demonstrate that LBR substantially alleviates length bias while consistently improving recommendation accuracy and fairness, with negligible additional training and inference overhead (with an average NDCG@5 gain of 16.82%). The code is available at https://github.com/Void-JackLee/LBR.
Hongchen Li, Bohao Wang, Jingbang Chen et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
