2026· Annual Meeting of the Association for Computational Linguistics· pp. 30754-30768· 0 citations· 39 references
Computer Science
TL;DR
This paper proposes LoRE, a novel two-stage training framework for e-commerce search relevance, which outperforms GPT-5 by 29.1% in Macro-F1 and achieving a relative 27% online gain, offering a vital reference for industrial domain-specific post-training post-training.
Abstract
E-commerce search relevance is a critical component of retrieval systems. While Large Language Models (LLMs)-driven Chain-of-Thought (CoT) modeling has become the dominant paradigm and yielded significant gains, a critical gap remains: the absence of a systematic definition for comprehensive relevance reasoning, which leads to significant blind spots in current approaches. In this paper, we de-construct the task into three core competencies: reasoning & knowledge, multi-modal understanding, and rule awareness. Accordingly, we propose LoRE ( L arge Generative M o del for Search R elevanc e ), a novel two-stage training framework. We first employ an SFT phase to instill these capabilities via a progressive CoT synthesis pipeline, followed by a Reinforcement Learning (RL) phase, which serves as a regularizer, pruning redundant logic to achieve precise and robust adjudication. Extensive experiments validate LoRE, outperforming GPT-5 by 29.1% in Macro-F1 and achieving a relative 27% online gain, offering a vital reference for industrial domain-specific post-training
E-commerce search relevance modeling faces a critical dilemma: traditional models falter with complex queries, while Large Language Models (LLMs), despite their superior reasoning, suffer from the prohibitive latency of auto-regressive Chain-of-Thought (CoT) generation, rendering them infeasible for production. Knowledge distillation offers a promising solution, yet current methods force an undesirable trade-off: sacrificing the very interpretability that makes LLMs powerful, or relying on expensive, unscalable human-annotated rationales. To address this, we propose STAR—Staged Training with Aligned Reinforcement Learning and Multi-Faceted Distillation, a progressive framework that follows a reasoning, ranking, and transfer pipeline to imbue dense models with both high performance and interpretability. First, STAR aligns a teacher LLM's reasoning with task objectives using a novel multi-granularity reward in Group Relative Policy Optimization (GRPO), leveraging only binary labels. Next, it refines the teacher's ability for calibrated scoring via token-level supervision, enabling efficient ranking through a single forward pass without any additional layers. Finally, this ''white-box'' knowledge is transferred to a compact student via multi-faceted distillation that preserves both reasoning logic and ranking behavior. Offline experiments demonstrate that our 0.6B student model rivals the performance of a strong 8B baseline, making it highly efficient and fully deployable. Real-world effectiveness is validated by significant online A/B test gains, including a +0.93% GoodRate lift and a +1.04% increase in GMV. STAR has been fully deployed to 100% of main search traffic on 1688.com.
Chenxu Wang, Jianzhi Shao, Chi Zhang et al.· Annual International ACM SIG...· 0 citations
This work proposes ChainPrune, a novel reasoning path semantic structural optimization method to efficiently and controllably synthesize self-generated high-quality training data and incorporates a DPO-based preference learning method combined with supervised loss, effectively mitigating false reward suppression.
Weihang Pan, Zhengxu Yu, Yuxiang Zhang et al.· 1 citation
Large Language Models (LLMs) are powerful but limited by static parametric knowledge that becomes outdated once pretraining ends. Knowledge editing addresses this problem by updating model behavior on target facts without full retraining. In particular, in-context knowledge editing has gained attention because it is training-free and readily applicable to black-box LLMs. Recent reinforcement learning (RL)-based approaches improve over fixed retrieval strategies by adapting prompt construction to the quantity-quality trade-off. Despite initial success, they fail to model the prompt as a structured entity under the distinct and often competing objectives of reliability, generality, and specificity. Previous methods largely optimize a single objective and make decisions over only part of the prompt construction process, thereby overlooking both the balance of different objectives and the global organization of demonstrations. We propose Multi-Objective In-context Knowledge Editing (MO-IKE), a multi-objective RL algorithm that formulates prompt construction for in-context knowledge editing as a Constrained Markov Decision Process. MO-IKE trains a dynamic retriever to optimize competing objectives in knowledge editing, enabling more balanced and globally coherent prompt construction. On Llama-3.2, MO-IKE improves edit success (reliability) from 85.0% to 92.0%, paraphrase consistency (generality) from 77% to 79%, while increasing retention rate (specificity) by 23.0% compared to prior RL-based methods.
Xu-Zhong Wang, Maiqi Jiang, Tejal Nair et al.· 1 citation
In large-scale e-commerce search, accurately modeling multi-modal relevance is paramount for matching user intent—especially given the growing influence of visual content on shopping decisions. However, existing methods often fail to perform fine-grained reasoning. For instance, they struggle when a product title is irrelevant due to marketing language, but its image is highly relevant to the query. Furthermore, they cannot effectively disambiguate which specific query terms are satisfied by the visual versus the textual modality. While Large Language Models (LLMs) excel at such reasoning, their high computational overhead makes direct online deployment infeasible. To bridge this gap, we propose TRACE (Term-level Reasoning And Chain-of-thought Enhanced distillation), a framework designed for deploying advanced reasoning capabilities at scale. TRACE operates in two stages. First, it enhances an LLM's multi-modal reasoning by employing Group Relative Policy Optimization (GRPO) guided by a term-level Chain-of-Thought (CoT) reward function, enabling it to generate detailed, step-by-step relevance judgments. Second, it efficiently transfers this fine-grained reasoning to a lightweight, deployable model using a novel term-level knowledge distillation strategy that inherits reasoning ability. Offline evaluations show significant improvements across different datasets. More critically, online A/B tests on 1688.com resulted in a +1.04% GMV uplift, a +0.906% LTV increase, and a +0.523% improvement in UV_L2O, demonstrating its significant value in a real-world production environment.
Chenxu Wang, Chi Zhang, Fang liang et al.· Annual International ACM SIG...· 0 citations
Large Language Models (LLMs) demonstrate significant potential in sequential recommendation, and leveraging their Chain-of-Thought (CoT) reasoning capabilities can further unlock profound user preference understanding. However, deploying explicit CoT reasoning in real-world systems faces prohibitive challenges: (i) the conflict between the large model scale required for high-fidelity reasoning and the resource constraints of online services, and (ii) the excessive latency introduced by auto-regressive rationale generation. To address these issues, we propose I Reasoning via Multi-Teacher Distillation (IRMD), a novel framework that 'compiles' the reasoning abilities of large teacher LLMs into a lightweight student Small Language Model (SLM). IRMD first employs a Multi-Teacher CoT Synthesis with Dual-Constraint Rejection Sampling module to generate a high-quality, diverse set of reasoning paths. Subsequently, our Annealing-Scheduled Reasoning Distillation strategy progressively trains the student to internalize this logic, transitioning from mimicking explicit CoT to performing purely implicit reasoning. Extensive experiments on multiple benchmark datasets demonstrate that IRMD significantly outperforms state-of-the-art baselines in both recommendation accuracy and inference efficiency. Our code is accessible at https://github.com/Cxx-0/IRMD.
Guided Retrieval Training (GRT) is introduced, a novel method that improves the performance of a search agent by restricting the retrieval process during RL training using ground truth information, and enhances training efficiency by achieving better QA performance with fewer training steps.
Aounon Kumar, Sudipta Paul, Vivek Kulkarni et al.· 0 citations