Title: Web Retrieval Agents for Evidence-Based Misinformation Detection Show Chinese title

Title: 基于证据的虚假信息检测的网络检索代理

Abstract: This paper develops an agent-based automated fact-checking approach for detecting misinformation. We demonstrate that combining a powerful LLM agent, which does not have access to the internet for searches, with an online web search agent yields better results than when each tool is used independently. Our approach is robust across multiple models, outperforming alternatives and increasing the macro F1 of misinformation detection by as much as 20 percent compared to LLMs without search. We also conduct extensive analyses on the sources our system leverages and their biases, decisions in the construction of the system like the search tool and the knowledge base, the type of evidence needed and its impact on the results, and other parts of the overall process. By combining strong performance with in-depth understanding, we hope to provide building blocks for future search-enabled misinformation mitigation systems. Show Chinese abstract

Abstract: 本文开发了一种基于代理的自动化事实核查方法，用于检测错误信息。 我们证明了将一个强大的大型语言模型代理（该代理无法进行网络搜索）与在线网页搜索代理相结合，其效果优于单独使用每个工具。 我们的方法在多个模型中都具有鲁棒性，优于其他方法，并且与没有搜索功能的大型语言模型相比，将错误信息检测的宏观F1分数提高了多达20%。 我们还对系统所利用的来源及其偏差进行了深入分析，包括系统构建中的决策，如搜索工具和知识库的选择、所需证据的类型及其对结果的影响，以及其他整体过程的部分。 通过结合出色的表现与深入的理解，我们希望为未来的搜索增强型错误信息缓解系统提供基础组件。