Pith. sign in

REVIEW 3 major objections 4 minor 58 references

MV-Bench: Benchmarking Multimodal Large Language Models for Coordinated Multi-View Interface Construction

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Current AI models can copy how a dashboard looks, but they cannot wire up the data and cross-view interactions that make it work.

desk verdict Useful new benchmark for coordinated multi-view interface generation, but the headline numbers come from a mislabeled subset and the reference ground truth is only verified visually, not semantically. read the letter →

arxiv 2607.19910 v1 pith:XSDQV7T3 submitted 2026-07-22 cs.CV cs.HC

Yue Zhao , Hongxu Liu , Feiyu Wang , Xiaoyu Yang , Tong Ge , Zhen Yang , Chao Wang , Qiong Zeng This is my paper
classification cs.CVcs.HC
keywords multimodallargelanguagemodelscoordinatedmultipleviewsbenchmarkimage-to-codegenerationdatabindinginteractioncompletenessTableauvisualizationcode
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that today's multimodal large language models, when asked to turn a screenshot of a coordinated multi-view dashboard into working code, can reproduce how the dashboard looks but largely fail at the two things that make it functional: binding the right data values to each view, and implementing the cross-view interactions (filtering, highlighting, brushing) that link views together. To test this, the authors built MV-Bench, a benchmark of 1,048 verified dashboard instances generated from Tableau workbook files through a pipeline that converts structured specifications into executable web code with data and interaction annotations. On the strongest model, visual fidelity reached 75.45%, while data binding scored 21.71% and interaction completeness 11.68%; the gap persisted even after up to three rounds of error-driven repair. The paper argues this is a capability limit, not a single-attempt artifact, and that visual appearance is an unreliable proxy for functional correctness in MLLM-generated interfaces.

What carries the argument

The benchmark's engine is a multi-stage pipeline that turns Tableau workbook XML into executable web applications via two structured intermediate representations: the Tableau Structured Specification (TSS), which normalizes authoring intent, and the Code Translation Specification (CTS), which resolves rendering commitments into execution directives. Reference code is produced by an LLM agent and accepted only after source, build, and render validation; the 92 accepted base interfaces are then recombined across chart types, datasets, and interaction patterns under interaction-preservation rules to yield 1,048 verified instances. Evaluation decomposes into three automatic metrics—Static (visua

What would settle it

Take a random subset of the reference interfaces, have a human expert verify the data bindings and interaction behavior against the original Tableau workbooks, and re-run the interaction episodes; if a substantial fraction of reference interfaces fail their own interaction tests, the scores would no longer support the capability-limit claim. Conversely, if a model that is explicitly given the CTS specification rather than only the screenshot achieves high Data and Int. scores, the gap may be a prompting artifact rather than a capability limit.

Watch

Extended reading notes

Core claim

Using Tableau workbook files as ground truth because they explicitly encode data bindings and interactions, MV-Bench provides an image-to-code benchmark for coordinated multi-view interface construction. The central finding is a consistent, large separation between static visual fidelity and functional correctness across all five evaluated models: every model scores much higher on reproducing the appearance of the reference image than on matching its data values or replaying its interaction episodes, and this separation survives iterative repair. The paper attributes the gap to representation asymmetry—layout is visible in the input image while data semantics and interaction logic are latent

Load-bearing premise

The benchmark's ground-truth interfaces are generated by an LLM and validated only for whether the code builds and renders, not for whether the data bindings and interaction behavior are actually correct, so the Data and Int. scores rest on the assumption that this reference generation preserves the original Tableau semantics.

Editorial extensions

If this is right

  • Any tool that evaluates or ranks MLLM code generation for dashboards using only screenshots will systematically overstate capability; correctness must be verified by executing the generated code and checking data and behavior.
  • The visual-versus-functional gap, persistent across models and repair rounds, indicates that simply scaling current models or giving them more attempts will not close the interaction and data-binding hole; new training objectives or architectures that model shared state across views are needed.
  • MV-Bench's replayable interaction episodes and sandboxed execution can serve as a feedback signal for iterative generation, letting a repair loop target data-binding and interaction failures rather than just compile errors.
  • Because the benchmark is built from Tableau dashboards with a fixed React/TypeScript/D3 stack, its results describe this stack; retargeting the pipeline to declarative grammars is left as future work but would test whether the gap is stack-specific.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the benchmark's ground truth is itself LLM-generated and only checked for source, build, and render validity, then the Data and Int. scores could partially measure agreement with an LLM's own errors rather than functional correctness; a hand-verified subset would settle this.
  • The same pipeline could be turned into a training signal: because data-binding and interaction failures are localized in the code, a future system could use MV-Bench's DOM-level checks as reward signals for reinforcement learning on cross-view reasoning.
  • The paper's framing suggests a concrete testable extension: evaluate MLLMs on declarative targets such as Vega-Lite where views, selections, and links are compositional—if the gap persists there, it points to a reasoning deficit rather than an imperative-coding artifact.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper introduces MV-Bench, a benchmark for evaluating multimodal large language models (MLLMs) on generating coordinated multi-view interfaces from screenshots, data, and interaction specifications. Tableau workbooks from the DMiner corpus are converted via two intermediate representations (TSS, CTS) into executable React/TypeScript/D3 code. The benchmark claims 92 base interfaces and 956 derived samples (1,048 total). Five state-of-the-art MLLMs are evaluated under single-pass and repair conditions with metrics for static visual fidelity, data binding correctness, and interaction completeness. The headline finding is a consistent gap: models achieve 67.9–76.4% static fidelity but only 0.9–21.7% data binding and 0–11.7% interaction completeness. The paper argues this gap reflects a fundamental capability limitation rather than an artifact of the single-pass setting, since three repair rounds raise executability but not the functional scores.

Significance. If the benchmark and its ground truth are valid, MV-Bench fills a real gap: existing image-to-code evaluations largely ignore data semantics and cross-view coordination. The metrics go beyond visual similarity by extracting rendered data values and replaying interaction episodes, and the pipeline from Tableau workbooks to executable code is a useful contribution. The consistent visual-vs-functional gap across five models and both benchmark partitions is an important empirical observation for the visualization and MLLM communities. However, the significance is conditional on two issues that are load-bearing for the quantitative claims: the reported results appear to come from a 103-instance subset rather than the claimed 1,048 instances, and the reference ground truth is validated only at the source/build/render level, not for data-binding or interaction semantics.

major comments (3)
  1. [Sec. 5 / Table 1; Sec. 3.5.3] The paper states the benchmark comprises 92 base interfaces and 956 derived samples (1,048 total, Sec. 3.5.3), but Table 1 reports evaluation on '72 Base Interfaces, 31 Derived Samples' and labels the combined 103 instances as 'Full Benchmark.' The headline numbers (e.g., Kimi 75.45% Static, 21.71% Data, 11.68% Int.) therefore rest on a subset containing 78% of base interfaces but only 3.2% of derived samples, not on the full 1,048 instances. No selection protocol for this subset is described. The paper must either evaluate on the full benchmark or clearly rename Table 1 as a subset and revise the abstract, conclusion, and all 'full benchmark' claims accordingly.
  2. [Sec. 3.4.2, 3.6.1, Eqs. (3)-(4)] Reference implementations are generated by GLM 4.7 and accepted after three checks: source (CTS fields referenced, data file loads), build (lint/type/build pass), and render (views present and non-empty). None of these verifies that rendered data values are semantically correct relative to the original Tableau workbook, nor that interaction actions produce correct cross-view effects. The only fidelity check against the original Tableau dashboards is visual (SSIM 0.7042, Sec. 3.5.3). Consequently, Sdata and Sint measure agreement with an LLM-generated reference artifact, not functional correctness. This circularity risk is not acknowledged in the Limitations section. The authors should add independent validation of reference data bindings and interaction behavior (e.g., human audit or automated extraction from .twb) and report error rates.
  3. [Sec. 5 intro; Sec. 5.6] The 'representative 10% subset' is unbalanced: 72/92 base (78%) and 31/956 derived (3.2%). The aggregate Full Benchmark scores average these 72+31 instances per interface, giving base interfaces about 70% weight, whereas the full benchmark has only ~8.8% base. Because Data and Int. scores differ substantially between partitions (e.g., Kimi Data: 20.04% base vs. 36.05% derived), the subset composition can materially change aggregate results. The selection protocol (random seed, stratification criteria) must be specified, and results should be reported with per-partition weighting or on the full 1,048 instances.
minor comments (4)
  1. [Sec. 4.2, Eq. (3)] The Data metric should specify how DOM-extracted values are matched (tolerances for numerical values, handling of axis scales or coordinate transforms). Without this detail, the metric is not fully reproducible.
  2. [Sec. 5 / Table 1] Consider reporting confidence intervals or bootstrap estimates for the reported scores, especially given the small subset and the acknowledged API nondeterminism (Sec. 6.2).
  3. [Sec. 3.5.3] The average SSIM of 0.7042 between reference implementations and original Tableau dashboards indicates nontrivial visual divergence. The paper should discuss whether this limits the use of these references as ground truth for Static scores.
  4. [Abstract / Conclusion] The abstract and conclusion repeat the 1,048-instance claim without noting that the evaluation reported in Table 1 was performed on a 103-instance subset. These statements need to be reconciled with the revised Table 1 presentation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Data/Int. gaps are empirical measurements against Tableau-derived references, not reductions to fitted inputs or self-citations.

full rationale

MV-Bench's derivation chain is not circular. The reference interfaces U are produced from Tableau workbooks via TSS/CTS (Sec. 3.3) and then used as comparison targets in Eqs. 3-4; the evaluated models are not used to construct U, and no fitted parameter is renamed as a prediction. The central claim (Sec. 5.3/6.1) is an empirical result: models score 67.9-76.4% Static but 0.9-21.7% Data and 0-11.7% Int., and the repair condition (Sec. 5.4) reports the same gap. The closest concern is the reference-generation pipeline: Sec. 3.6.1 states all references are generated by GLM 4.7, and Sec. 3.4.2's source/build/render checks do not verify data-binding semantics or interaction behavior. This is a genuine external-validity limitation—especially since Sec. 6.2 does not flag the possibility that references inherit data/coordination errors—but it is not circularity: low Data/Int. scores mean models fail to match even this LLM-generated reference, and the qualitative error analysis (fabricated values, omitted/wrong interaction bindings, Sec. 5.5) gives independent evidence for the gap. The fact that GLM 4.7 generated references while GLM 4.6V is evaluated does not force the result, since four other models are also evaluated and could have scored high if the references were trivially reproducible. Self-citations (e.g., Zhao et al. [53-55]) appear only in Related Work and are not load-bearing. No equation reduces to its own input by construction, so the circularity score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central result rests on treating Tableau workbooks as trustworthy specifications, on the sufficiency of automatic verification for ground truth, and on the representativeness of the evaluation subset. TSS and CTS are internal file formats, not explanatory postulates, so no invented entities carry explanatory burden.

assumptions (4)
  • domain assumption Tableau workbook files (.twb/.twbx) encode reliable ground truth for data bindings, visual mappings, and interaction definitions.
    Used as the ground-truth source throughout construction (Sec. 3.2.1). If the XML omits or mis-specifies semantics, MV-Bench inherits those errors.
  • ad hoc to paper The three-level verification pipeline (source/build/render) is sufficient to certify correctness of reference implementations.
    Sec. 3.4.2: render validation only checks that all expected views are present and non-empty; it does not verify data-binding values or interaction behavior, so errors can pass into the ground truth.
  • domain assumption The 103-instance evaluation subset is representative of the full 1,048-instance benchmark.
    Sec. 5 states 'representative 10% subset' but the base:derived ratio is 72:31 (70:30) versus 92:956 (9:91) in the full benchmark; no stratified sampling procedure is given.
  • domain assumption The fixed React/TypeScript/D3 target stack does not fundamentally change the visual-vs-functional gap.
    The authors themselves note in Sec. 6.2(6) that the imperative stack may make data binding and interaction harder than declarative alternatives, so the gap may be stack-specific rather than a general MLLM limitation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MV-Bench: Benchmarking Multimodal Large Language Models for Coordinated Multi-View Interface Construction." pith.science (2026). https://pith.science/paper/XSDQV7T3

@misc{pith2026260719910,
  author       = {Pith},
  title        = {Pith review of: MV-Bench: Benchmarking Multimodal Large Language Models for Coordinated Multi-View Interface Construction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XSDQV7T3}},
  note         = {Machine review of arXiv:2607.19910}
}
read the original abstract

Multimodal large language models (MLLMs) are increasingly expected to automate visualization development by generating code directly from visual designs. However, existing evaluations mainly focus on single-chart generation and overlook coordinated multi-view interface construction, which requires joint reasoning about data semantics, view coordination, and interaction logic. Consequently, MLLM capabilities in this setting remain underexplored, and the field lacks a dedicated benchmark for systematic assessment. We introduce MV-Bench, a benchmark for evaluating MLLMs on coordinated multi-view interface construction. Instead of relying on incomplete or inconsistent open-source implementations, we use Tableau workbook files as ground truth because they explicitly encode data bindings, visual mappings, and interactions. We develop a multi-stage pipeline that converts these specifications into executable web interfaces through structured intermediate representations. The benchmark contains 92 base interfaces and 1,048 verified instances created by recombining chart types, datasets, and interaction patterns. Each instance includes executable code, a rendered interface, a dataset, and interaction annotations. We evaluate five state-of-the-art MLLMs in a single-pass setting using metrics for visual fidelity, data binding correctness, and interaction completeness. The strongest model achieves 75.45 percent accuracy in visual layout reproduction, but only 21.71 percent in data binding and 11.68 percent in interaction completeness. These results show that current MLLMs can reproduce visual appearance but remain limited in generating the data semantics and interactive logic required by coordinated multi-view interfaces. Iterative refinement improves code executability but does not substantially reduce the gap in data binding and interaction generation.

Figures

Figures reproduced from arXiv: 2607.19910 by the authors.

Figure 1
Figure 1. MV-Bench benchmarks multimodal code generation for coordinated multi-view interface construction. (a) Constructing interfaces from visual designs requires recovering data bindings, layout structures, and interaction logic from heterogeneous and often incomplete specifications. (b) MV-Bench transforms these fragmented workbook specifications into structured intermediate representations, and generates executable bench… view at source ↗
Figure 2
Figure 2. Construction pipeline of MV-Bench. We parse Tableau work￾books into structured intermediate representations–Tableau Structured Specification (TSS) and Code Translation Specification (CTS), gener￾ate executable web interfaces, and verify correctness at different levels. Derived samples are produced through controlled recomposition with interaction-preservation constraints. 3 MV-BENCH CONSTRUCTION 3.1 Overview We aim … view at source ↗
Figure 3
Figure 3. Example of distributed interaction semantics in a Tableau work [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Overview of MV-Bench. The benchmark contains 92 base interfaces and 956 derived samples. Example dashboards illustrate variation in multi-view layout, chart combinations, and interaction patterns. 4 EVALUATION METHODOLOGY We use MV-Bench to evaluate how well current ML…
Figure 5
Figure 5. Figure 5: Performance across chart types (bar, line, scatter, pie, and table). [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Representative failure modes of MLLM-generated interfaces, including incorrect layout and chart composition, incorrect data binding, incorrect [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

58 extracted references · 6 canonical work pages

  1. [1]

    R. Anil, S. Borgeaud, et al. Gemini: A family of highly capable multimodal models.arXiv preprint arXiv:2312.11805, 2023. doi: 10.48550/arXiv. 2312.11805 7

    arXiv doi:10.48550/arxiv.2312.11805 2023

  2. [2]

    Anthropic. Claude. 2024.https://www.anthropic.com/. 2, 7

    2024

  3. [3]

    J. Bai, S. Bai, Y . Chu, Z. Cui, K. Dang, X. Deng, Y . Fan, W. Ge, Y . Han, F. Huang, et al. Qwen technical report.arXiv preprint arXiv:2309.16609,

    arXiv

  4. [4]

    T. Bai, Y . Bai, Y . Bao, S. Cai, Y . Cao, Y . Charles, H. Che, C. Chen, G. Chen, et al. Kimi k2.5: Visual agentic intelligence.arXiv preprint arXiv:2602.02276, 2026. doi: 10.48550/arXiv.2602.02276 2, 7

    arXiv doi:10.48550/arxiv.2602.02276 2026

  5. [5]

    Beltramelli

    T. Beltramelli. pix2code: Generating code from a graphical user interface screenshot. InProceedings of the ACM SIGCHI Symposium on Engi- neering Interactive Computing Systems, EICS ’18, article no. 3, 6 pages. Association for Computing Machinery, New York, NY , USA, 2018. doi: 10.1145/3220134.3220135 3

    arXiv 2018

  6. [6]

    Bostock, V

    M. Bostock, V . Ogievetsky, and J. Heer. D3: Data-driven documents.IEEE Transactions on Visualization and Computer Graphics, 17(12):2301–2309,

  7. [8]

    X. Chen, L. Gong, A. Cheung, and D. Song. PlotCoder: Hierarchical decoding for synthesizing visualization code in programmatic context. InProceedings of the 59th Annual Meeting of the Association for Com- putational Linguistics and the 11th International Joint Conference on Natural Language Processing, vol. 1, pp. 2169–2181. Association for Computational ...

    doi:10.18653/v1/2021.acl-long 2021

  8. [9]

    Y . Chen, R. Li, A. Mac, T. Xie, T. Yu, and E. Wu. Nl2interface: Interactive visualization interface generation from natural language queries.IEEE Visualization Conference NLVIZ Workshop 2022. doi: 10.48550/arXiv. 2209.08834 3

    work page Pith review arXiv doi:10.48550/arxiv.2209.08834 2022

Show all 58 references
  1. [10]

    B. Deka, Z. Huang, C. Franzen, J. Hibschman, D. Afergan, Y . Li, J. Nichols, and R. Kumar. Rico: A mobile app dataset for building data-driven design applications. InProceedings of the 30th Annual ACM Symposium on User Interface Software and Technology, UIST ’17, 10 pages, p. ...

    2017

  2. [11]

    V . Dibia. LIDA: A tool for automatic generation of grammar-agnostic visu- alizations and infographics using large language models. InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics, pp. 113–126. Association for Computational Linguistics...

  3. [12]

    Y . Gao, X. Hu, T. Peng, and H. Jiang. A rule-based approach for UI migration from Android to iOS.arXiv preprint arXiv:2409.16656, 2024. doi: 10.48550/arXiv.2409.16656 3

    2024 doi arXiv

  4. [13]

    T. Ge, B. Lee, and Y . Wang. Cast: Authoring data-driven chart animations. InProceedings of the 2021 CHI Conference on Human Factors in Comput- ing Systems, CHI ’21, article no. 24, 15 pages. Association for Computing Machinery, New York, NY , USA, 2021. doi: 10.1145/3411764.3445452 2

    2021

  5. [14]

    doi: 10.18653/v1/2023.acl-demo.11 1

    2023 doi

  6. [15]

    Goswami, P

    K. Goswami, P. Mathur, R. Rossi, and F. Dernoncourt. Plotgen: Multi- agent llm-based scientific data visualization via multimodal retrieval feed- back. InCompanion Proceedings of the ACM on Web Conference 2025, WWW ’25, 5 pages, p. 1672–1676. Association for Computing Machiner...

    2025

  7. [16]

    Y . Han, C. Zhang, X. Chen, et al. Chartllama: A multimodal llm for chart understanding and generation.arXiv preprint arXiv:2311.16483, 2023. doi: 10.48550/arXiv.2311.16483 2

    2023 doi arXiv

  8. [17]

    L. Gong, C. Wang, D. Cui, Y . Huang, and M. Wei. Uitrans: Seamless ui translation from android to harmonyos. InProceedings of the 16th International Conference on Internetware, Internetware ’25, 5 pages, p. 142–146. Association for Computing Machinery, New York, NY , USA,

  9. [18]

    C. E. Jimenez, J. Yang, A. Wettig, S. Yao, K. Pei, O. Press, and K. Narasimhan. SWE-bench: Can language models resolve real-world GitHub issues?arXiv preprint arXiv:2310.06770, 2024. doi: 10.48550/ arXiv.2310.06770 3

    2024 doi arXiv

  10. [19]

    D. Keim, G. Andrienko, J.-D. Fekete, C. Görg, J. Kohlhammer, and G. Melançon.Visual Analytics: Definition, Process, and Challenges, pp. 154–175. Springer Berlin Heidelberg, Berlin, Heidelberg, 2008. doi: 10. 1007/978-3-540-70956-5_7 3, 6

    2008

  11. [20]

    Laurençon, L

    H. Laurençon, L. Tronchon, and V . Sanh. Unlocking the conversion of web screenshots into HTML code with the WebSight dataset.arXiv preprint arXiv:2403.09029, 2024. doi: 10.48550/arXiv.2403.09029 3

    2024 doi arXiv

  12. [21]

    Heer and B

    J. Heer and B. Shneiderman. Interactive dynamics for visual analysis. Commun. ACM, 55(4):45–54, 10 pages, 2012. doi: 10.1145/2133806. 2133821 2

    2012 doi

  13. [22]

    R. Li, Y . Zhang, and D. Yang. Sketch2Code: Evaluating vision-language models for interactive web design prototyping. InProceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies, vol....

    2025 doi

  14. [23]

    S. Li, X. Chen, Y . Song, Y . Song, C. J. Zhang, F. Hao, and L. Chen. Prompt4vis: Prompting large language models with example mining for tabular data visualization.The VLDB Journal, 34(4):38, 2025. doi: 10. 1007/s00778-025-00912-0 3

    2025

  15. [25]

    D. Li, H. Mei, Y . Shen, S. Su, W. Zhang, J. Wang, M. Zu, and W. Chen. Echarts: A declarative framework for rapid construction of web-based visualization.Visual Informatics, 2(2):136–146, 2018. doi: 10.1016/j. visinf.2018.04.011 2

    2018 doi

  16. [26]

    Y . Luo, N. Tang, G. Li, C. Chai, W. Li, and X. Qin. Synthesizing natural language to visualization (nl2vis) benchmarks from nl2sql benchmarks. InProceedings of the 2021 International Conference on Management of Data, SIGMOD ’21, 13 pages, p. 1235–1247. Association for Computi...

    2021

  17. [27]

    Y . Luo, N. Tang, G. Li, J. Tang, C. Chai, and X. Qin. Natural language to visualization by neural machine translation.IEEE Transactions on Visualization and Computer Graphics, 28(1):217–226, 2022. doi: 10. 1109/TVCG.2021.3114848 2

    2022

  18. [28]

    Masry, D

    A. Masry, D. X. Long, J. Q. Tan, S. Joty, and E. Hoque. ChartQA: A benchmark for question answering about charts with visual and logical reasoning. InFindings of the Association for Computational Linguistics: ACL 2022, pp. 2263–2279. Association for Computational Linguistics, ...

    2022 doi

  19. [29]

    Z. Liu, J. Thompson, A. Wilson, M. Dontcheva, J. Delorey, S. Grigg, B. Kerr, and J. Stasko. Data illustrator: Augmenting vector design tools with lazy data binding for expressive visualization authoring. InProceed- ings of the 2018 CHI Conference on Human Factors in Computing ...

    2018

  20. [30]

    Munzner.Visualization Analysis and Design

    T. Munzner.Visualization Analysis and Design. AK Peters Visualization Series. CRC Press, 2015. doi: 10.1145/3721241.3733989 2

    2015

  21. [31]

    T. Niu, Y . Cui, B. Wang, X. Xu, X. Yao, Q. Zhu, D. Wu, S. Wang, and W. Che. Chart2Code53: A large-scale diverse and complex dataset for en- hancing chart-to-code generation. InProceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pp. 15828–15...

    2025

  22. [32]

    GPT-4o system card.arXiv preprint arXiv:2410.21276, 2024

    OpenAI. GPT-4o system card.arXiv preprint arXiv:2410.21276, 2024. doi: 10.48550/arXiv.2410.21276 1, 2, 7

    2024 doi arXiv

  23. [33]

    Moreira, L

    G. Moreira, L. Ferreira, C. Veiga, M. Hosseini, and F. Miranda. Urban- ite: A dataflow-based framework for human-ai interactive alignment in urban visual analytics.IEEE Transactions on Visualization and Computer Graphics, 32(1):1065–1075, 2026. doi: 10.1109/TVCG.2025.3634644 3

    2026

  24. [34]

    D. Ren, B. Lee, and M. Brehmer. Charticulator: Interactive construction of bespoke chart layouts.IEEE Transactions on Visualization and Computer Graphics, 25(1):789–799, 2019. doi: 10.1109/TVCG.2018.2865158 2

    2019

  25. [35]

    J. C. Roberts. State of the art: Coordinated & multiple views in exploratory visualization. InFifth International Conference on Coordinated and Multiple Views in Exploratory Visualization (CMV 2007), pp. 61–71, 2007. doi: 10.1109/CMV.2007.20 2, 3

    2007 doi

  26. [36]

    Satyanarayan, D

    A. Satyanarayan, D. Moritz, K. Wongsuphasawat, and J. Heer. Vega-lite: A grammar of interactive graphics.IEEE Transactions on Visualization and Computer Graphics, 23(1):341–350, 2017. doi: 10.1109/TVCG.2016. 2599030 2

    2017 doi

  27. [37]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever. Learning transferable visual models from natural language supervision. In Proceedings of the 38th International Conference on Machine Lea...

    2021

  28. [38]

    S. Shi, T. Ren, G. Zhu, G. Feng, and J. Hu. Closing the feedback loop in text2vis: Refining visualization with vision-language models. InProceed- ings of the 33rd ACM International Conference on Multimedia, MM ’25, 9 pages, p. 9053–9061. Association for Computing Machinery, Ne...

    2025

  29. [39]

    C. Si, Y . Zhang, R. Li, Z. Yang, R. Liu, and D. Yang. Design2Code: Bench- marking multimodal code generation for automated front-end engineering. InProceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Huma...

    2025 doi

  30. [40]

    Sievert.Interactive Web-Based Data Visualization with R, Plotly, and Shiny

    C. Sievert.Interactive Web-Based Data Visualization with R, Plotly, and Shiny. Chapman & Hall/CRC, 2020. doi: 10.1201/9780429447273 2

    2020 doi

  31. [41]

    L. Shen, E. Shen, Y . Luo, X. Yang, X. Hu, X. Zhang, Z. Tai, and J. Wang. Towards natural language interfaces for data visualization: A survey.IEEE Transactions on Visualization and Computer Graphics, 29(6):3121–3144, 24 pages, 2023. doi: 10.1109/TVCG.2022.3148007 2

    2023

  32. [42]

    P.-P. Vázquez. Are llms ready for visualization? In2024 IEEE 17th Pacific Visualization Conference (PacificVis), pp. 343–352, 2024. doi: 10. 1109/PacificVis60374.2024.00049 2

    2024

  33. [43]

    C. Wang, J. Thompson, and B. Lee. Data formulator: Ai-powered concept- driven visualization authoring.IEEE Transactions on Visualization and Computer Graphics, 30(1):1128–1138, 2024. doi: 10.1109/TVCG.2023. 3326585 3

    2024 doi

  34. [44]

    H. W. Wang, M. Gordon, L. Battle, and J. Heer. Dracogpt: Extracting visualization design preferences from large language models.IEEE Trans- actions on Visualization and Computer Graphics, 31(1):710–720, 2025. doi: 10.1109/TVCG.2024.3456350 3

    2025

  35. [45]

    Y . Tian, W. Cui, D. Deng, X. Yi, Y . Yang, H. Zhang, and Y . Wu. Chartgpt: Leveraging llms to generate charts from abstract natural language.IEEE Transactions on Visualization and Computer Graphics, 31(3):1731–1745,

  36. [46]

    doi: 10.1109/TVCG.2024.3368621 2, 3

    2024

  37. [47]

    R. Xia, H. Ye, X. Yan, et al. Chartx & chartvlm: A versatile benchmark and foundation model for complicated chart reasoning.IEEE Transactions on Image Processing, 2025. 3

    2025

  38. [48]

    L. Xie, C. Zheng, H. Xia, H. Qu, and C. Zhu-Tian. Waitgpt: Monitoring and steering conversational llm agent in data analysis with on-the-fly code visualization. InProceedings of the 37th Annual ACM Symposium on User Interface Software and Technology, UIST ’24, article no. 119,...

    2024

  39. [49]

    Xu et al

    Z. Xu et al. ChartBench: A benchmark for complex visual reasoning in charts.arXiv preprint arXiv:2312.15915, 2024. doi: 10.48550/arXiv.2312 .15915 2

    2024 doi arXiv

  40. [50]

    Y . Wang, J. Pan, D. Shi, et al. Chartoptimiser: Task-driven optimisation of chart designs.arXiv preprint arXiv:2504.10180, 2025. doi: 10.48550/ arXiv.2504.10180 1

    2025 doi

  41. [51]

    M. O. Ward, G. Grinstein, and D. A. Keim.Interactive Data Visualization: Foundations, Techniques, and Applications. A K Peters / CRC Press, Boca Raton, FL, USA, 2010. doi: 10.1201/9780429108433 2

    2010 doi

  42. [52]

    H. H. Zhang, T. Zhang, B. Lin, Y . Xue, Y . Zhu, H. Liu, L. Gu, L. Ye, Z. Wang, X. Zuo, et al. Widget2code: From visual widgets to ui code via multimodal llms.arXiv preprint arXiv:2512.19918, 2025. doi: 10. 48550/arXiv.2512.19918 3

    2025 doi

  43. [53]

    Y . Zhao, J. Wang, L. Xiang, X. Zhang, Z. Guo, C. Turkay, Y . Zhang, and S. Chen. Lightva: Lightweight visual analytics with llm agent-based task planning and execution.IEEE Transactions on Visualization and Computer Graphics, 31(9):6162–6177, 2025. doi: 10.1109/TVCG.2024. 3496112 3

    2025 doi

  44. [54]

    Y . Zhao, Y . Wang, X. Luo, Y . Wang, and J.-D. Fekete. Libra: An in- teraction model for data visualization. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, CHI ’25, article no. 1169, 17 pages. Association for Computing Machinery, New York, NY ...

    2025

  45. [55]

    Z. Yang, Z. Zhou, S. Wang, X. Cong, X. Han, Y . Yan, Z. Liu, Z. Tan, P. Liu, D. Yu, Z. Liu, X. Shi, and M. Sun. MatPlotAgent: Method and evaluation for LLM-based agentic scientific data visualization. InFindings of the Association for Computational Linguistics: ACL 2024, pp. 1...

    2024 doi

  46. [56]

    A. Zeng, B. Xu, B. Wang, C. Zhang, D. Yin, D. Zhang, D. Rojas, G. Feng, H. Zhao, et al. Chatglm: A family of large language models from glm- 130b to glm-4 all tools.arXiv preprint arXiv:2406.12793, 2024. doi: 10. 48550/arXiv.2406.12793 2, 7

    2024 doi arXiv

  47. [60]

    Y . Zhao, Y . Zhang, Y . Zhang, X. Zhao, J. Wang, Z. Shao, C. Turkay, and S. Chen. Leva: Using large language models to enhance visual analytics. IEEE Transactions on Visualization and Computer Graphics, 31(3):1830– 1847, 2025. doi: 10.1109/TVCG.2024.3368060 3

    2025

  48. [2011]

    doi: 10.1109/TVCG.2011.185 2

    2011 doi

  49. [2023]

    doi: 10.48550/arXiv.2309.16609 2, 7

    doi arXiv

  50. [2025]

    doi: 10.1145/3755881.3755933 3

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.