# EvidenceLens: A Claim-Evidence Matrix for Auditing Financial Question Answering

arXiv: https://arxiv.org/abs/2606.23724 · HTML: https://mlsanity.com/paper/2606.23724/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 26 references that are in the map.
- First posted on arXiv 19 Jun 2026.

## Builds on (most cited first, 19 of 19)

- [ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning](https://mlsanity.com/paper/2203.10244/) (2022)
- [FinQA: A Dataset of Numerical Reasoning over Financial Data](https://mlsanity.com/paper/2109.00122/) (2021)
- [TAT-QA: A Question Answering Benchmark on a Hybrid of Tabular and Textual Content in Finance](https://mlsanity.com/paper/2105.07624/) (2021)
- [How Good (Or Bad) Are LLMs at Detecting Misleading Visualizations?](https://mlsanity.com/paper/2407.17291/) (2024)
- [FinSage: A Multi-aspect RAG System for Financial Filings Question Answering](https://mlsanity.com/paper/2504.14493/) (2025)
- [Unraveling the Truth: Do VLMs really Understand Charts? A Deep Dive into Consistency and Robustness](https://mlsanity.com/paper/2407.11229/) (2024)
- [Characterizing Automated Data Insights](https://mlsanity.com/paper/2008.13060/) (2020)
- [How Aligned are Human Chart Takeaways and LLM Predictions? A Case Study on Bar Charts with Varying Layouts](https://mlsanity.com/paper/2408.06837/) (2024)
- [WikiMixQA: A Multimodal Benchmark for Question Answering over Tables and Charts](https://mlsanity.com/paper/2506.15594/) (2025)
- [The Perils of Chart Deception: How Misleading Visualizations Affect Vision-Language Models](https://mlsanity.com/paper/2508.09716/) (2025)
- [SimVecVis: A Dataset for Enhancing MLLMs in Visualization Understanding](https://mlsanity.com/paper/2506.21319/) (2025)
- [Understanding Why ChatGPT Outperforms Humans in Visualization Design Advice](https://mlsanity.com/paper/2508.01547/) (2025)
- [Enhancing Large Vision-Language Models with Layout Modality for Table Question Answering on Japanese Annual Securities Reports](https://mlsanity.com/paper/2505.17625/) (2025)
- [Does visualization help AI understand data?](https://mlsanity.com/paper/2507.18022/) (2025)
- [Enhancing XAI Interpretation through a Reverse Mapping from Insights to Visualizations](https://mlsanity.com/paper/2508.18640/) (2025)
- [Reflection on Data Storytelling Tools in the Generative AI Era from the Human-AI Collaboration Perspective](https://mlsanity.com/paper/2503.02631/) (2025)
- [Challenges & Opportunities with LLM-Assisted Visualization Retargeting](https://mlsanity.com/paper/2507.01436/) (2025)
- [ReVise: A Human-AI Interface for Incremental Algorithmic Recourse](https://mlsanity.com/paper/2508.00002/) (2025)
- [Toward a Logic of Generalization about Visualization as a Decision Aid](https://mlsanity.com/paper/2508.06751/) (2025)

## Abstract (arXiv)

Large language models are increasingly used to answer questions over annual reports, earnings decks, and analyst notes, yet their outputs remain difficult to verify in high-stakes financial workflows. A fluent answer can blend directly grounded statements, weak synthesis, and unsupported claims across narrative text, tables, and charts. We present EvidenceLens, a visual analytics prototype that treats financial question answering as a claim-evidence alignment problem. The system decomposes an answer into atomic claims, summarizes support composition and confidence, support gaps, and coordinates claim-level inspection with source passages, table cells, and chart regions. Its core visual representation is a multimodal claim-evidence matrix that makes coverage, contradiction, and modality imbalance immediately visible. To support reproducibility, we also specify a JSON-based artifact schema, a lightweight multimodal alignment pipeline, and a deterministic review-priority ranking that maps backend signals into an auditable visual structure. Through representative report-auditing scenarios, we show how EvidenceLens helps analysts distinguish grounded claims from overconfident synthesis that conventional chat interfaces flatten.