Robust Denoising Neural Reranker for Recommender Systems

2025-09-23

AI Summary

In multi-stage recommendation, existing re-ranking modules underutilize retrieval scores and are vulnerable to noise. To address this, we propose the Denoising Neural Re-ranking (DNR) framework, the first to formulate re-ranking as a robust denoising process over retrieval scores. DNR introduces an adversarial noise generation mechanism and jointly optimizes three objectives: (i) score denoising, (ii) adversarial reconstruction of retrieval scores, and (iii) distribution regularization—thereby enhancing robustness and generalization. Built upon deep neural networks, DNR precisely aligns user feedback with denoised retrieval scores. Extensive experiments on three public benchmarks demonstrate that DNR significantly outperforms state-of-the-art re-ranking methods in both effectiveness and efficiency. Theoretical analysis ensures soundness, while architectural design supports practical deployment in industrial recommender systems.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Data Mining & Knowledge Management: Recommender Systems
• Search and Optimization: Distributed Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

For multi-stage recommenders in industry, a user request would first trigger a simple and efficient retriever module that selects and ranks a list of relevant items, then calls a slower but more sophisticated deep reranking model that refines the item arrangement before exposure to the user. The latter model typically reranks the item list conditioned on the user's history content and the initial ranking from retrievers. Although this two-stage retrieval-ranking framework demonstrates practical effectiveness, the significance of retriever scores from the previous stage has been limitedly explored, which is informative. In this work, we first theoretically analyze the limitations of using retriever scores as the rerankers' input directly and argue that the reranking task is essentially a noise reduction problem from the retriever scores. Following this notion, we derive an adversarial framework, DNR, that associates the denoising reranker with a carefully designed noise generation module. We extend the conventional score error minimization term with three augmented objectives, including: 1) a denoising objective that aims to denoise the noisy retriever scores to align with the user feedback; 2) an adversarial retriever score generation objective that improves the exploration in the retriever score space; and 3) a distribution regularization term that aims to align the distribution of generated noisy retriever scores with the real ones. Extensive experiments are conducted on three public datasets, together with analytical support, validating the effectiveness of the proposed DNR.

Problem

• Addresses limitations of using retriever scores directly in reranking systems
• Treats reranking as a noise reduction problem from retriever scores
• Proposes adversarial framework to denoise scores and align with user feedback

Innovation

• Adversarial framework DNR for denoising retriever scores
• Three augmented objectives to refine reranking process
• Aligns generated noisy scores with real user feedback