Elamur Research⌕
☰
home / papers / 2026 / RADAR: A Reasoning-Guided Attribution Framework for Explainable Visual Data Analysis

RADAR: A Reasoning-Guided Attribution Framework for Explainable Visual Data Analysis

Paper · 2026 · Anku Rani, Aparna Garimella, Apoorv Saxena et al.

Summary

Multimodal LLMs can answer questions about charts, but they do not show which visual elements led to their answers. This makes their output hard to verify in areas such as business, medicine or policy. RADAR addresses this with attribution for mathematical chart QA at two levels. It marks bounding boxes for the final answer and also for each intermediate reasoning step.

The authors built a benchmark from ChartQA covering bar and line charts. Claude 3.5 Sonnet first drafted reasoning chains and box annotations. Human annotators then checked and corrected them in three stages: reasoning validation, answer attribution and step attribution.

The proposed method first generates reasoning steps and then uses them to guide region attribution. It relies on hidden states of InternLM-XComposer2 and a sliding-window procedure. Attribution quality is checked automatically by regenerating answers from the highlighted regions.

Compared with baselines, using reasoning as guidance raises attribution accuracy by as much as 15 percentage points. Answers produced from the attributions reach BERTScore around 0.90 against the ground truth. The work points toward chart-analysis systems whose conclusions users can trace and verify.

Key points

  • Benchmark size: 1000 ChartQA charts, 2000 question-answer pairs, 3599 reasoning steps and 11,220 attribution boxes.
  • Annotation process: two annotators hired via Upwork worked about 120 hours at $15/hour using the VGG Image Annotator. Inter-annotator agreement on the attribution stages was only 52–64%.
  • Pilot study (100 ChartQA items, human-judged):
    • Answer accuracy: Claude 3.5 Sonnet 96%, GPT-4v 64%, GPT-4o 58%.
    • Reasoning accuracy: Claude 3.5 Sonnet 75%, GPT-4o 49%, GPT-4v 45%.
    • Giving models the gold answer cut reasoning failures from 51% to 25%.
  • Model details: the attribution model is InternLM-XComposer2, using layer-16 hidden states. It beats GPT-4o-style attribution by up to 15 pp, and regenerated answers reach BERTScore of about 0.90.
  • Generalization: a fully automatic run on 1068 ChartQA pie charts gave BERTScore of about 0.9 and STS of about 0.5. Inference takes roughly 1–2 seconds per chart-question pair.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 3The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksComputer Vision and Pattern Recognition · 2,869 works

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 1,426 works

Cite

Cite as Anku Rani, Aparna Garimella, Apoorv Saxena, Balaji Vasan Srinivasan and Paul Pu Liang. 2026. RADAR: A Reasoning-Guided Attribution Framework for Explainable Visual Data Analysis. Findings of the Association for Computational Linguistics: EACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-eacl.299

@inproceedings{rani2026radar,
  title = {{RADAR: A Reasoning-Guided Attribution Framework for Explainable Visual Data Analysis}},
  author = {Anku Rani, Aparna Garimella, Apoorv Saxena, Balaji Vasan Srinivasan and Paul Pu Liang},
  booktitle = {Findings of the Association for Computational Linguistics: EACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-eacl.299},
  url = {https://doi.org/10.18653/v1/2026.findings-eacl.299}
}

References

29 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Eneko Agirre, Daniel Cer, Mona Diab, and Aitor Gonzalez-Agirre. 2012. SemEval-2012 task 6: A pilot on semantic textual similarity. In *SEM 2012: The First Joint Conference on Lexical and Computational Semantics – Volume 1: Proceedings of the main conference and the shared task, and Volume 2: Proceed…
  2. Janice Ahn, Rishu Verma, Renze Lou, Di Liu, Rui Zhang, and Wenpeng Yin. 2024. Large language models for mathematical reasoning: Progresses and challenges. arXiv preprint arXiv:2402.00157.
  3. Bernd Bohnet, Vinh Q Tran, Pat Verga, Roee Aharoni, Daniel Andor, Livio Baldini Soares, Massimiliano Ciaramita, Jacob Eisenstein, Kuzman Ganchev, Jonathan Herzig, et al. 2022. Attributed question answering: Evaluation and modeling for attributed large language models. arXiv preprint arXiv:2212.08037…
  4. Zheng Cai, Maosong Cao, Haojiong Chen, Kai Chen, Keyu Chen, Xin Chen, Xun Chen, Zehui Chen, Zhi Chen, Pei Chu, Xiaoyi Dong, Haodong Duan, Qi Fan, Zhaoye Fei, Yang Gao, Jiaye Ge, Chenya Gu, Yuzhe Gu, Tao Gui, Aijia Guo, Qipeng Guo, Conghui He, Yingfan Hu, Ting Huang, Tao Jiang, Penglong Jiao, Zhenjia… — InternLM2 Technical Report
  5. Jacob Cohen. 1960. A coefficient of agreement for nominal scales. Educational and psychological measurement, 20(1):37–46. find
  6. Xiaoyi Dong, Pan Zhang, Yuhang Zang, Yuhang Cao, Bin Wang, Linke Ouyang, Xilin Wei, Songyang Zhang, Haodong Duan, Maosong Cao, et al. 2024. Internlm-xcomposer2: Mastering free-form text-image composition and comprehension in vision-language large model. arXiv preprint arXiv:2401.16420.
  7. Himanshu Gupta, Shreyas Verma, Ujjwala Anantheswaran, Kevin Scaria, Mihir Parmar, Swaroop Mishra, and Chitta Baral. 2024. Polymath: A challenging multi-modal mathematical reasoning benchmark. find
  8. Yucheng Han, Chi Zhang, Xin Chen, Xu Yang, Zhibin Wang, Gang Yu, Bin Fu, and Hanwang Zhang. 2023. Chartllama: A multimodal llm for chart understanding and generation. arXiv preprint arXiv:2311.16483. find
  9. Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685.
  10. Shima Imani, Liang Du, and Harsh Shrivastava. 2023. Mathprompter: Mathematical reasoning using large language models. arXiv preprint arXiv:2303.05398.
  11. Mohammed Saidul Islam, Raian Rahman, Ahmed Masry, Md Tahmid Rahman Laskar, Mir Tafseer Nayeem, and Enamul Hoque. 2024. Are large vision language models up to the challenge of chart comprehension and reasoning? an extensive investigation into the capabilities and limitations of lvlms. arXiv preprint… find
  12. Dae Hyun Kim, Enamul Hoque, and Maneesh Agrawala. 2020. Answering questions about charts and generating visual explanations. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, CHI ’20, page 1–13, New York, NY, USA. Association for Computing Machinery. find
  13. Pan Lu, Liang Qiu, Wenhao Yu, Sean Welleck, and Kai-Wei Chang. 2023. A survey of deep learning for mathematical reasoning. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 14605– 14631, Toronto, Canada. Association for Computat… find
  14. Junyu Luo, Zekun Li, Jinpeng Wang, and Chin-Yew Lin. 2021. Chartocr: Data extraction from charts images via a deep hybrid framework. In Proceedings of the IEEE/CVF winter conference on applications of computer vision, pages 1917–1925. find
  15. Ahmed Masry, Do Long, Jia Qing Tan, Shafiq Joty, and Enamul Hoque. 2022. ChartQA: A benchmark for question answering about charts with visual and logical reasoning. In Findings of the Association for Computational Linguistics: ACL 2022, pages 2263– 2279, Dublin, Ireland. Association for Computationa…
  16. Puneet Mathur, Alexa Siu, Nedim Lipka, and Tong Sun. 2024. MATSA: Multi-agent table structure attribution. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 250–258, Miami, Florida, USA. Association for Computational Linguistics. find
  17. Anirudh Phukan, Divyansh, Harshit Kumar Morj, Vaishnavi, Apoorv Saxena, and Koustava Goswami. 2024a. Beyond logit lens: Contextual embeddings for robust hallucination detection & grounding in vlms. find
  18. Anirudh Phukan, Shwetha Somasundaram, Apoorv Saxena, Koustava Goswami, and Balaji Vasan Srinivasan. 2024b. Peering into the mind of language models: An approach for attribution in contextual question answering. In Findings of the Association for Computational Linguistics ACL 2024, pages 11481–11495,… find
  19. Jirui Qi, Gabriele Sarti, Raquel Fernández, and Arianna Bisazza. 2024. Model internals-based answer attribution for trustworthy retrieval-augmented generation. arXiv preprint arXiv:2406.13663. supervision. In International Conference on Machine Learning. find
  20. Leonardo Ranaldi and Andre Freitas. 2024. Aligning large and small language models via chain-of-thought reasoning. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1812–1827, St. Julian’s, Malta. Association…
  21. Hamid Rezatofighi, Nathan Tsoi, JunYoung Gwak, Amir Sadeghian, Ian Reid, and Silvio Savarese. 2019. Generalized intersection over union: A metric and a loss for bounding box regression. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 658–666. find
  22. Ankit Satpute, Noah Gießing, André Greiner-Petter, Moritz Schubotz, Olaf Teschke, Akiko Aizawa, and Bela Gipp. 2024. Can llms master math? investigating large language models on math stack exchange. In Proceedings of the 47th International ACM SI-GIR Conference on Research and Development in Informa… find
  23. Pragya Srivastava, Manuj Malik, Vivek Gupta, Tanuja Ganu, and Dan Roth. 2024. Evaluating llms’ mathematical reasoning in financial document question answering. In Findings of the Association for Computational Linguistics ACL 2024, pages 3853–3878. find
  24. Manan Suri, Puneet Mathur, Nedim Lipka, Franck Dernoncourt, Ryan A. Rossi, and Dinesh Manocha. 2025. ChartLens: Fine-grained visual attribution in charts. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 22447–22462, Vienna, Au… find
  25. Zhi Rui Tam, Cheng-Kuang Wu, Yi-Lin Tsai, Chieh-Yen Lin, Hung yi Lee, and Yun-Nung Chen. 2024. Let me speak freely? a study on the impact of format restrictions on performance of large language models. find
  26. Hanchen Wang, Tianfan Fu, Yuanqi Du, Wenhao Gao, Kexin Huang, Ziming Liu, Payal Chandak, Shengchao Liu, Peter Van Katwyk, Andreea Deac, et al. 2023. Scientific discovery in the age of artificial intelligence. Nature, 620(7972):47–60. find
  27. Xiang Yue, Boshi Wang, Ziru Chen, Kai Zhang, Yu Su, and Huan Sun. 2023. Automatic evaluation of attribution by large language models. arXiv preprint arXiv:2305.06311. find
  28. Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger, and Yoav Artzi. 2019. Bertscore: Evaluating text generation with bert. ArXiv, abs/1904.09675. find
  29. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. Learning transferable visual models from natural language find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/rani-2026-radar-reasoning-guided-attribution-framework/