让表格问答的每一步推理都有据可查，提升可信度。
TraceBack: Multi-Agent Decomposition for Fine-Grained Table Attribution
- 用多智能体分解问题，精准定位支持答案的单元格。
- 在多个数据集上显著优于基线，细粒度归因准确率更高。
- 适合需要高可信度的医疗、金融等关键领域应用。
结构化表格上的问答不仅需要准确答案，还需明确指出支撑答案的具体单元格。现有表格问答系统很少提供细粒度归因，导致即使答案正确也缺乏可验证依据，限制了其在高风险场景中的信任度。为此，我们提出TraceBack，一个模块化多智能体框架，实现单表问答中可扩展的单元格级归因。TraceBack通过剪枝相关行与列、将问题分解为语义连贯的子问题，并将每个答案片段与其支持的单元格对齐，捕捉中间推理步骤中的显性和隐性证据。为支持系统评估，我们发布CITEBench基准，包含来自ToTTo、FetaQA和AITQA的短语-单元格标注。我们还提出FairScore，一种无需参考标签的指标，通过比较预测单元格与答案导出的原子事实，估算归因精度与召回率。实验表明，TraceBack在多个数据集和粒度下显著超越强基线，FairScore与人工判断高度一致，且保持方法间的相对排名，支持可解释且可扩展的表格问答评估。
原文摘要 · Abstract (English)
Question answering (QA) over structured tables requires not only accurate answers but also transparency about which cells support them. Existing table QA systems rarely provide fine-grained attribution, so even correct answers often lack verifiable grounding, limiting trust in high-stakes settings. We address this with TraceBack, a modular multi-agent framework for scalable, cell-level attribution in single-table QA. TraceBack prunes tables to relevant rows and columns, decomposes questions into semantically coherent sub-questions, and aligns each answer span with its supporting cells, capturing both explicit and implicit evidence used in intermediate reasoning steps. To enable systematic evaluation, we release CITEBench, a benchmark with phrase-to-cell annotations drawn from ToTTo, FetaQA, and AITQA. We further propose FairScore, a reference-less metric that compares atomic facts derived from predicted cells and answers to estimate attribution precision and recall without human cell labels. Experiments show that TraceBack substantially outperforms strong baselines across datasets and granularities, while FairScore closely tracks human judgments and preserves relative method rankings, supporting interpretable and scalable evaluation of table-based QA.
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。