面对矛盾信息，大模型答不准？新数据集揭示问题并提出解释引导训练法。
Open Domain Question Answering with Conflicting Contexts
- 构建人工标注数据集QACC，发现25%开放域问题检索出矛盾上下文。
- 三大主流大模型在矛盾上下文中准确率不足，暴露推理短板。
- 通过让模型生成答案解释，提升其处理信息冲突的推理能力。
开放域问答系统常依赖大规模文本（如网络）获取信息，但这些文本常含矛盾内容，盲目使用会导致错误答案。为评估该问题严重性，我们构建了人工标注数据集QACC，发现高达25%的明确开放域问题在使用Google搜索时会产生矛盾上下文。我们在此数据集上评测并基准化三个强大的大语言模型（LLMs），揭示其在处理矛盾信息时的局限性。为探究人类如何推理矛盾信息，我们要求标注者提供正确答案的选择依据。结果显示，通过微调模型生成答案解释，可引入更丰富的推理信息，引导模型有效应对矛盾上下文。
原文摘要 · Abstract (English)
Open domain question answering systems frequently rely on information retrieved from large collections of text (such as the Web) to answer questions. However, such collections of text often contain conflicting information, and indiscriminately depending on this information may result in untruthful and inaccurate answers. To understand the gravity of this problem, we collect a human-annotated dataset, Question Answering with Conflicting Contexts (QACC), and find that as much as 25% of unambiguous, open domain questions can lead to conflicting contexts when retrieved using Google Search. We evaluate and benchmark three powerful Large Language Models (LLMs) with our dataset QACC and demonstrate their limitations in effectively addressing questions with conflicting information. To explore how humans reason through conflicting contexts, we request our annotators to provide explanations for their selections of correct answers. We demonstrate that by finetuning LLMs to explain their answers, we can introduce richer information into their training that guide them through the process of reasoning with conflicting contexts.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。