UniRank: Unified List-wise Reranking via Confidence-Ordered Denoising

TL;DR

UniRank is a novel unified list-wise reranking framework that combines autoregressive and non-autoregressive paradigms through an iterative denoising process, improving performance in various datasets and real-world applications.

Contribution

It introduces a unified architecture that merges AR and NAR rerankers, utilizing bidirectional modeling and denoising, with a new Task Grounded Diffusion Interface for item-level denoising.

Findings

Outperforms state-of-the-art baselines on multiple datasets.

Achieves significant online A/B test improvements in user engagement metrics.

Unifies AR and NAR reranking with a novel iterative denoising approach.

Abstract

List-wise reranking arranges a request-specific pool of candidate items into an ordered slate that maximizes user satisfaction. Existing generative rerankers fall into two paradigms: Autoregressive (AR) rerankers construct the slate left to right and capture inter-item dependencies in the exposure list, but they suffer from error propagation because early mistakes affect subsequent slots. Non-autoregressive (NAR) rerankers predict all slots in parallel and avoid error propagation, but they weaken inter-item interaction modeling under a slot independence assumption. This raises a central question: is there a unified architecture that combines the strengths of both paradigms and delivers stronger reranking performance? We answer this question with UniRank, a unified list-wise reranking framework whose inference time variants recover AR and NAR rerankers as special cases. UniRank integrates bidirectional slate modeling into an iterative denoising process and fills the most confident slot at each step. To instantiate this framework for reranking, we introduce the Task Grounded Diffusion Interface (TGD), which performs denoising at the item level and restricts prediction to the request-specific candidate pool. TGD aggregates each item's semantic tokens into a single item embedding and scores each slot directly against the candidate pool. Experiments on Amazon Books, MovieLens-1M, and an industrial short video dataset show that UniRank consistently outperforms state-of-the-art baselines. Online A/B tests on a real-world industrial platform further validate its effectiveness, yielding significant improvements of +0.159% in user average app-time and +1.016% in share-rate.