TRACE: Agentic Catalog Enrichment with Multi-source Evidence Grounding

📄 arXiv: 2608.20844v1 📥 PDF

作者: Rohan Kumar, Steven Xu, Kyle MacDonald, Matthew Long, Bernice Chow, Mac VanRenterghem, Sudeep Das

分类: cs.AI

发布日期: 2026-08-21

备注: 12 pages, 2 figures


💡 一句话要点

提出TRACE框架以解决电商目录属性稀缺问题

🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)

关键词: 电商目录 属性丰富 大型语言模型 多模态证据 自动化 用户体验 数据验证

📋 核心要点

  1. 现有电商目录往往属性稀缺,导致消费者和系统无法有效利用信息,手动丰富目录不切实际。
  2. TRACE框架利用代理大型语言模型自动化丰富目录属性,通过多模态证据验证候选属性值。
  3. 实验结果显示,TRACE在离线评估中达到98.2%的准确率,并在实际应用中提升了90.4%的属性覆盖率。

📝 摘要(中文)

产品目录在电商中是搜索、发现和推荐的基础,但往往属性稀缺,消费者和下游系统依赖的属性要么埋藏在标题和图片等非结构化内容中,要么在目录中完全缺失。手动丰富电商目录在规模和快速增长的情况下是不切实际的。本文提出TRACE,一个基于代理的大型语言模型(LLM)的自动化目录属性丰富框架。ScoutAgent通过商家目录、联合数据源和身份匹配的网络搜索三角测量多模态证据,提出候选属性值及其支持证据,而JudgeAgent则验证每个属性值的支持证据,并决定是否发布或转交人工审核。在离线人类评估数据集上,TRACE提出的属性值准确率为98.2%,属性覆盖率为74.7%。在行业规模的目录中部署后,TRACE在四个业务领域的印象加权丰富覆盖率提高了90.4%。在线实验显示,产品详情页上展示丰富的属性使结账转化率提高了0.48%。

🔬 方法详解

问题定义:本文旨在解决电商产品目录中属性稀缺的问题。现有方法通常依赖于手动更新,难以应对目录的快速增长和规模。

核心思路:TRACE框架通过引入代理大型语言模型,自动化地从多种数据源中提取和验证产品属性,减少人工干预,提高效率。

技术框架:TRACE由两个主要模块组成:ScoutAgent和JudgeAgent。ScoutAgent负责收集和分析多模态证据,提出候选属性值;JudgeAgent则对候选值进行验证,确保其准确性和可靠性。

关键创新:TRACE的创新在于结合了多模态证据和代理模型的能力,使得属性提取和验证过程更加高效和准确,显著提升了属性覆盖率。

关键设计:在设计中,TRACE采用了特定的损失函数来优化属性值的验证过程,并通过多模态数据融合技术提高了模型的整体性能。

🖼️ 关键图片

fig_0
fig_1

📊 实验亮点

TRACE在离线评估中实现了98.2%的属性值准确率和74.7%的属性覆盖率。在实际应用中,印象加权丰富覆盖率提升了90.4%,并且在线实验显示,丰富的属性展示使结账转化率提高了0.48%。

🎯 应用场景

TRACE框架在电商领域具有广泛的应用潜力,能够自动化地丰富产品目录属性,提高用户搜索和推荐的准确性,进而提升消费者的购物体验。未来,该技术还可以扩展到其他需要信息丰富的领域,如内容推荐和广告投放等。

📄 摘要(原文)

Product catalogs underpin search, discovery, and recommendation in e-commerce, yet they are often attribute-sparse: the attributes shoppers and downstream systems rely on are either buried in unstructured content such as titles and images or missing from the catalog altogether. Manually enriching e-commerce catalogs is impractical given their scale and rapid growth. This paper introduces TRACE, a novel framework for automated catalog attribute enrichment using agentic Large Language Models (LLMs). A ScoutAgent triangulates multimodal evidence across merchant catalogs, syndicated feeds, and identity-matched web search to propose candidate attribute values with supporting evidence, while a JudgeAgent verifies the proposed value for each attribute value against its supporting evidence and decides whether to publish it or route it to human review. On an offline human evaluation dataset, TRACE's proposed attribute values were 98.2% accurate at 74.7% attribute coverage. Deployed in production on an industry-scale catalog, TRACE increased impression-weighted enrichment coverage across four business verticals by 90.4%. An online experiment subsequently showed that surfacing the enriched attributes on the product detail page increased checkout conversion by 0.48%.