Pith. sign in

REVIEW 4 major objections 4 minor 39 references

SimVecVis: A Dataset for Enhancing MLLMs in Visualization Understanding

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Fine-tuning on SimVecVis lifts MLLM chart-data accuracy from 12% to 54%.

desk verdict A useful synthetic chart dataset and a plausible fine-tuning recipe, but the headline accuracy numbers are not yet reliable due to an undocumented evaluation protocol and an internally inconsistent table row. read the letter →

arxiv 2506.21319 v3 pith:QOVXOC5X submitted 2025-06-26 cs.HC cs.CV

Can Liu , Chunlin Da , Xiaoxiao Long , Yuxiao Yang , Yu Zhang , Yong Wang This is my paper
classification cs.HCcs.CV
keywords visualizationunderstandingmultimodallargelanguagemodelschartquestionansweringchain-of-thoughtreasoningvectorrepresentationfine-tuningdatasetconstructionreconstruction
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

The paper claims that a new training dataset, SimVecVis, can substantially improve how multimodal large language models read data from charts. The dataset pairs 2,999 bitmap chart images with two extra supervision signals: a compact structured 'SimVec' description of the chart's marks and coordinates, and question-answer pairs whose answers are justified by explicit chain-of-thought reasoning anchored to axis scales. The headline result is that fine-tuning MiniCPM on SimVec plus QA-with-CoT raises data-question accuracy within 5% of ground truth from 11.92% zero-shot to 53.84%, outperforming zero-shot GPT-4o's 16.54%. The paper further shows that a model trained with SimVec can reconstruct charts from pixels at high fidelity, and that the recipe only helps models with strong spatial localization. If true, this gives a route to practical chart QA for scanned documents and historical visualizations, where data is not available as metadata.

What carries the argument

SimVec is a simplified vector format that flattens a chart into an ordered list of text, rectangle, line, and polygon elements, each with normalized coordinates in a 0–1000 space and HSL colors quantized to a 0–20 range. It is compact, reducing token count by about 90% compared to the SVG version of the same chart, and it is machine-readable enough to serve both as training supervision and as a reconstruction target. The other load-bearing component is the chain-of-thought QA annotation, which anchors each reasoning step to ground-truth axis metadata, for example: the Y-axis maps 50 pixels to 450 pixels corresponding to 0% to 100%, the bar height is 140 pixels, so the value is (140/(450−50))×100 = 35%. Together, these two components train the model to decode visual marks into data values rather than merely matching image patterns.

What would settle it

Run the fine-tuned MiniCPM model from Table 2 on a held-out set of real-world charts with non-linear axes, such as logarithmic scales, or with heavily distorted hand-drawn axes, and measure accuracy within 5% of ground truth; if it collapses toward the 11.92% zero-shot baseline, the learned pixel-to-value reasoning is an artifact of the linear template rather than a general chart-reading skill.

Watch

Extended reading notes

Core claim

The paper's central claim is that fine-tuning a multimodal large language model on SimVecVis—where each chart image is paired with a compact SimVec encoding and chain-of-thought QA traces—substantially improves data-centric chart question answering. The strongest demonstration is with MiniCPM: with SimVec plus QA-with-CoT supervision, 53.84% of predictions fall within 5% of the ground-truth value, compared with 11.92% for the zero-shot model and 16.54% for zero-shot GPT-4o. The same training does not help Qwen-VL, which the authors attribute to its weaker chart-element localization. The paper also shows that a model trained to output SimVec can reconstruct bar, line, and area charts from pixels, with text hit rates near 99% and average element position errors around 1% of image size on the evaluated cases, while noting that errors in early reasoning steps such as axis decoding still propagate to final answers.

Load-bearing premise

The chain-of-thought annotations are generated from ground-truth axis metadata that assumes a known, linear pixel-to-value mapping for each chart, and the training recipe depends on that mapping being present and learnable; real charts without clean, linear axes may not transfer the learned reasoning.

Editorial extensions

If this is right

  • Fine-tuning a spatially capable MLLM on SimVecVis with SimVec and chain-of-thought supervision yields large accuracy gains over zero-shot models on data-centric chart questions.
  • Chain-of-thought supervision alone improves accuracy over direct-answer training, and adding SimVec predictions produces a further large jump for models that can localize chart elements.
  • The benefit is not model-agnostic: a model with weaker chart-element localization showed no improvement from SimVec, so the recipe depends on the base model's spatial perception.
  • A model trained with SimVec can reconstruct chart structure from pixels with high fidelity, suggesting that SimVec is an expressive intermediate representation for chart understanding.
  • The main remaining failure mode is error accumulation in multi-step reasoning, such as an incorrect axis reading propagating to the final numerical answer.

Reading between the lines

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

  • Beyond the stated results, the linear axis-mapping assumption in the chain-of-thought annotations implies that the gains may not transfer to charts with logarithmic scales, broken axes, or other non-linear pixel-to-value mappings; that is a concrete stress test for the method.
  • If transfer to uncontrolled charts does hold, model-generated SimVec reconstructions could serve as pseudo-labels for unlabeled chart corpora, turning the explicit vector representation into a self-training signal.
  • The high reconstruction fidelity suggests SimVec could also function as an intermediate target for chart-to-code or chart-to-data-table pipelines, not just for question answering.
  • Since the recipe helped only a model with strong localization, pairing SimVec supervision with a localization-focused auxiliary objective might extend the benefit to weaker visual-language models.
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

4 major / 4 minor

Summary. The paper introduces SimVec, a simplified vector representation for chart elements, and SimVecVis, a dataset of 2,999 generated bar, line, and area charts, each with a bitmap image, a SimVec representation, and data-centric QA pairs with chain-of-thought annotations. The authors fine-tune MiniCPM and Qwen-VL under several ablations and report large gains, e.g., MiniCPM with SimVec + QA w/ CoT reaches 53.84% within-5% accuracy versus 11.92% zero-shot (Table 2). They also report a reconstruction experiment in which a fine-tuned model generates SimVec from chart images (Section 5.3). The paper concludes that SimVecVis and CoT supervision substantially improve data-centric visualization understanding.

Significance. If validated, the dataset is a useful resource: it pairs raster charts with an intermediate structured representation and CoT reasoning, and the compact SimVec format is a sensible target for reconstruction. The paper also reports a concrete training recipe with modest compute. However, the current evidence does not yet establish generalization, because the evaluation protocol is undocumented and one table row is internally inconsistent; the result is plausible but under-evidenced. The central contribution (dataset plus representation) is worth pursuing, and most concerns are addressable with additional documentation and experiments.

major comments (4)
  1. [Section 5, Table 2] The MiniCPM (QA w/o CoT) row reports 26.92% for <5%, 41.92% for <10%, and 25.38% for <20%; since the <20% criterion is a superset of the <10% criterion, this ordering is impossible. This entry must be corrected and the whole table rechecked, because the headline comparison (SimVec + QA w/ CoT 53.84% vs. zero-shot 11.92%) depends on this table.
  2. [Section 5, Tables 2 and 3] The manuscript never defines the train/test split, the test-set size, or the selection procedure for evaluation instances. Several percentages are consistent with n=260, but this is not stated. Because all charts are generated from a small set of predefined templates with randomized colors and axis ranges (Section 4.1), high in-distribution accuracy could arise from template regularity rather than chart understanding; without a held-out split or an external benchmark, the reported gains do not establish generalization.
  3. [Section 4.1] The CoT annotations are synthesized from ground-truth axis metadata, e.g., 'the Y-axis ... maps from 50 pixels to 450 pixels, corresponding to ... 0% to 100%'. This assumes a known, linear pixel-to-data mapping during training. Since real charts must be decoded from noisy images, the paper should test whether the learned reasoning transfers, for example by evaluating on an external chart-QA benchmark or on charts whose pixel-to-axis mapping must be visually estimated.
  4. [Section 5.3] The reconstruction experiment uses MiniCPM (SimVec + QA w/ CoT), a model fine-tuned on SimVecVis, to generate SimVec from chart images. Interpreting high reconstruction quality as evidence of SimVec's expressiveness is partly circular because the model was trained to emit this exact format; report held-out or few-shot performance with a model that has not seen SimVec training data, or reframe H3 as a test of learnability rather than expressiveness.
minor comments (4)
  1. [Section 3, Table 1] The note that 'all coordinates and size are described using a uniform value where the size is set to 1000' is ambiguous; please state explicitly that all coordinates are relative to a normalized 1000-unit canvas.
  2. [Section 4] The relationship among 2,999 visualizations, 2,999 identification tasks, 5,642 extreme-value tasks, and the number of QA pairs used for training and evaluation should be clarified.
  3. [Section 5.2] No hyperparameters, number of epochs, learning rate, or data split are reported; one paragraph on the training setup would improve reproducibility.
  4. [Section 4.2] The term 'mocked historical visualizations' is informal; 'simulated' or 'synthesized' would be clearer.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor circularity: the H3 reconstruction experiment tests a model fine-tuned to emit SimVec, so it partly assumes SimVec's sufficiency; the central QA improvement remains an independent empirical comparison.

  1. fitted input called prediction [Section 5.3 (Hypothesis H3), supported by the training definition in Section 5.2]
    "To evaluate hypothesis H3, we assess the reconstruction capabilities using the SimVec format. We use MiniCPM (SimVec+CoT) to take an image as input and generate the corresponding SimVec as output. ... These results validate H3, confirming that the SimVec format effectively supports high-fidelity reconstruction of both textual and graphical elements."

    The evaluated model, MiniCPM (SimVec+CoT), was fine-tuned with the additional objective to 'predict the SimVec representation of the chart' (Section 5.2). The reconstruction test then asks that same model to output exactly the quantity it was supervised to produce—the image-to-SimVec mapping. High reconstruction fidelity is therefore primarily a check that the model learned its training target, not an independent demonstration that SimVec is an expressive or sufficient chart encoding. No non-SimVec-trained baseline is reported, and the 100 evaluation images are not stated to be held out from training, so the H3 validation partly assumes what it claims to show.

full rationale

The principal claim—that fine-tuning MLLMs on SimVecVis improves data-centric chart QA—is supported by a direct empirical comparison across model variants (Table 2, e.g., MiniCPM SimVec+QA w/ CoT 53.84% vs. 11.92% zero-shot at <5% tolerance). This comparison is not circular: the QA supervision and the evaluation metric are distinct, and the result is a measured outcome rather than an identity. The only load-bearing reduction I can exhibit is in the secondary H3 reconstruction experiment: the evaluated model was fine-tuned to output SimVec from chart images, and the reconstruction test then measures that same output; calling this a validation of SimVec's expressiveness partly assumes the representation's sufficiency rather than establishing it independently. No uniqueness theorem or load-bearing self-citation is invoked, and the CoT annotations, while derived from ground-truth axis metadata, are training data rather than a circular evaluation input. Concerns about undocumented train/test splits or the inconsistent Table 2 row are correctness/validity issues, not circularity, so they do not raise the circularity score.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The paper introduces SimVecVis as a new dataset and the SimVec format as a representation, but the central claims are empirical rather than derived from a mathematical model, so there are no fitted free parameters or invented physical entities. The main unstated assumptions concern the representativeness of synthetic charts and the correctness of axis metadata used to generate chain-of-thought annotations.

assumptions (3)
  • domain assumption SimVec conversion removes enough SVG detail that no data-relevant information is lost.
    Section 3 states SimVec flattens groups, standardizes coordinates, and removes styling; if some visible element needed for data reading is dropped, chart reconstruction and QA would be incomplete.
  • domain assumption The axis metadata used to build CoT traces gives a correct linear pixel-to-data mapping.
    Section 4.1 describes CoT examples that map pixel distances to percentages using axis metadata; real charts may use nonlinear scales, unknown origins, or occluding elements.
  • domain assumption Synthetic charts generated from GPT-4o topics and predefined templates are representative of the visualizations for which MLLM understanding should improve.
    Only bar, line, and area charts are included, and historical-style charts are mocked from digital charts rather than collected from real historical documents (Section 4.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of SimVecVis: A Dataset for Enhancing MLLMs in Visualization Understanding." pith.science (2026). https://pith.science/paper/QOVXOC5X

@misc{pith2026250621319,
  author       = {Pith},
  title        = {Pith review of: SimVecVis: A Dataset for Enhancing MLLMs in Visualization Understanding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QOVXOC5X}},
  note         = {Machine review of arXiv:2506.21319}
}
read the original abstract

Current multimodal large language models (MLLMs), while effective in natural image understanding, struggle with visualization understanding due to their inability to decode the data-to-visual mapping and extract structured information. To address these challenges, we propose SimVec, a novel simplified vector format that encodes chart elements such as mark type, position, and size. The effectiveness of SimVec is demonstrated by using MLLMs to reconstruct chart information from SimVec formats. Then, we build a new visualization dataset, SimVecVis, to enhance the performance of MLLMs in visualization understanding, which consists of three key dimensions: bitmap images of charts, their SimVec representations, and corresponding data-centric question-answering (QA) pairs with explanatory chain-of-thought (CoT) descriptions. We finetune state-of-the-art MLLMs (e.g., MiniCPM and Qwen-VL), using SimVecVis with different dataset dimensions. The experimental results show that it leads to substantial performance improvements of MLLMs with good spatial perception capabilities (e.g., MiniCPM) in data-centric QA tasks. Our dataset and source code are available at: https://github.com/VIDA-Lab/SimVecVis.

Figures

Figures reproduced from arXiv: 2506.21319 by the authors.

Figure 1
Figure 1. We convert the metadata into digital-format charts and [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Historical-style visualizations with paper-texture, hand [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. For each case (A, B, and C), the top panel displays the [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

39 extracted references · 13 canonical work pages

  1. [1]

    R. Amar, J. Eagan, and J. Stasko. Low-level components of analytic activity in information visualization. In Proc. IEEE InfoVis. Symp., pp. 111–117, 2005. doi: 10.1109/INFVIS.2005.1532136

    arXiv 2005

  2. [2]

    J. Bai, S. Bai, S. Yang, S. Wang, S. Tan, P. Wang, J. Lin, C. Zhou, and J. Zhou. Qwen-VL: A versatile vision-language model for un- derstanding, localization, text reading, and beyond. 2023. doi: 10. 48550/ARXIV.2308.12966

    arXiv doi:10.48550/arxiv.2308.12966 2023

  3. [3]

    Battle, P

    L. Battle, P. Duan, Z. Miranda, D. Mukusheva, R. Chang, and M. Stone- braker. Beagle: Automated extraction and interpretation of visualiza- tions from the web. In Proc. ACM Conf. Hum. Factors Comput. Syst., pp. 1–8, 2018. doi: 10.1145/3173574.3174168

    arXiv 2018

  4. [4]

    M. A. Borkin, A. A. V o, Z. Bylinskii, P. Isola, S. Sunkavalli, A. Oliva, and H. Pfister. What makes a visualization memorable? IEEE Trans. Vis. Comput. Graph., 19(12):2306–2315, 2013. doi: 10.1109/TVCG. 2013.234

    doi:10.1109/tvcg 2013

  5. [5]

    Bostock, V

    M. Bostock, V . Ogievetsky, and J. Heer. D3: Data-Driven Documents. IEEE Trans. Vis. Comput. Graph., 17(12):2301–2309, 2011. doi: 10. 1109/TVCG.2011.185

    work page 2011

  6. [6]

    C. Chen, R. Zhang, E. Koh, S. Kim, S. Cohen, and R. Rossi. Figure captioning with relation maps for reasoning. InProc. IEEE Winter Conf. Appl. Comput. Vis., pp. 1537–1545, 2020. doi: 10.1109/W ACV45572. 2020.9093592

    arXiv 2020

  7. [7]

    K. Cox, R. E. Grinter, S. L. Hibino, L. J. Jagadeesan, and D. Mantilla. A multi-modal natural language interface to an information visualization environment. Int. J. Speech Technol., 4(3-4):297–314, 2001. doi: 10. 1023/A:1011368926479

    work page 2001

  8. [8]

    D. Deng, Y . Wu, X. Shu, J. Wu, S. Fu, W. Cui, and Y . Wu. VisImages: A fine-grained expert-annotated visualization dataset. IEEE Trans. Vis. Comput. Graph., 29(7):3298–3311, 2023. doi: 10.1109/TVCG.2022. 3155440

    doi:10.1109/tvcg.2022 2023

Show all 39 references
  1. [9]

    E. Fast, B. Chen, J. Mendelsohn, J. Bassen, and M. S. Bernstein. Iris: A conversational agent for complex tasks. In Proc. ACM Conf. Hum. Factors Comput. Syst., 2018. doi: 10.1145/3173574.3174047

    2018

  2. [10]

    X. Fu, Y . Wang, H. Dong, W. Cui, and H. Zhang. Visualization assessment: A machine learning approach. In Proc. IEEE Vis. Conf., pp. 126–130, 2019. doi: 10.1109/VISUAL.2019.8933570

    2019

  3. [11]

    Hoque and M

    E. Hoque and M. Agrawala. Searching the visual style and structure of D3 visualizations. IEEE Trans. Vis. Comput. Graph., 26(1):1236–1245,

  4. [12]

    K. Hu, M. A. Bakker, S. Li, T. Kraska, and C. Hidalgo. VizML: A machine learning approach to visualization recommendation. In Proc. ACM Conf. Hum. Factors Comput. Syst. , pp. 1–12, 2019. doi: 10. 1145/3290605.3300358

    2019

  5. [13]

    K. Hu, S. Gaikwad, M. Hulsebos, M. A. Bakker, E. Zgraggen, C. Hi- dalgo, T. Kraska, G. Li, A. Satyanarayan, and C ¸. Demiralp. VizNet: Towards a large-scale visualization learning and benchmarking reposi- tory. In Proc. ACM Conf. Hum. Factors Comput. Syst., pp. 1–12, 2019. doi...

    2019

  6. [14]

    S. Hu, Y . Tu, et al. MiniCPM: Unveiling the potential of small language models with scalable training strategies. 2024. doi: 10.48550/ARXIV. 2404.06395

    2024 doi arXiv

  7. [15]

    Kafle, B

    K. Kafle, B. L. Price, S. Cohen, and C. Kanan. DVQA: understanding data visualizations via question answering. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5648–5656, 2018

    2018

  8. [16]

    S. E. Kahou, V . Michalski, A. Atkinson, ´A. K´ad´ar, A. Trischler, and Y . Bengio. FigureQA: An annotated figure dataset for visual reasoning. In Workshop Proceedings of International Conference on Learning Representations, 2018

    2018

  9. [17]

    S. E. Kahou, V . Michalski, A. Atkinson, ´A. K´ad´ar, A. Trischler, and Y . Bengio. FigureQA: An annotated figure dataset for visual reasoning. In Proc. Workshop Track Int. Conf. Learn. Represent., 2018. doi: 10. 48550/arXiv.1710.07300

    2018 doi arXiv

  10. [19]

    C. Lai, Z. Lin, R. Jiang, Y . Han, C. Liu, and X. Yuan. Automatic annotation synchronizing with textual description for visualization. In Proc. ACM Conf. Hum. Factors Comput. Syst., 2020. doi: 10.1145/ 3313831.3376443

    2020

  11. [20]

    C. Liu, Y . Han, R. Jiang, and X. Yuan. ADVISor: Automatic visual- ization answer for natural-language question on tabular data. In Proc. IEEE Pac. Vis. Symp., pp. 6–15, 2021. doi: 10.1109/PacificVis52677. 2021.00010

    2021

  12. [21]

    C. Liu, R. Jiang, S. Tan, J. Yu, C. Yang, H. Shao, and X. Yuan. Datasets of visualization for machine learning, 2024. doi: 10.48550/arXiv.2407. 16351

    2024 doi

  13. [22]

    C. Liu, L. Xie, Y . Han, X. Yuan, et al. AutoCaption: An approach to generate natural language description from visualization automatically. In Proc. IEEE Pac. Vis. Symp. , pp. 191–195, 2020. doi: 10.1109/ PacificVis48177.2020.1043

    2020

  14. [23]

    H. Lu, W. Liu, B. Zhang, B. Wang, K. Dong, B. Liu, et al. DeepSeek- VL: Towards real-world vision-language understanding, 2024. doi: 10. 48550/arXiv.2403.05525

    2024 doi arXiv

  15. [24]

    Y . Luo, X. Qin, N. Tang, and G. Li. DeepEye: Towards automatic data visualization. In Proc. IEEE Int. Conf. Data Eng., pp. 101–112, 2018. doi: 10.1109/ICDE.2018.00019

    2018

  16. [25]

    Mackinlay, P

    J. Mackinlay, P. Hanrahan, and C. Stolte. Show me: Automatic presen- tation for visual analysis. IEEE Trans. Vis. Comp. Graph., 13(6):1137– 1144, 2007. doi: 10.1109/TVCG.2007.70594

    2007

  17. [26]

    X. Mei, Y . Zhang, C. Yang, R. Shi, and X. Yuan. ZuantuSet: A collection of historical chinese visualizations and illustrations. In Proc. ACM Conf. Hum. Factors Comput. Syst., 2025

    2025

  18. [27]

    Moritz, C

    D. Moritz, C. Wang, G. L. Nelson, H. Lin, A. M. Smith, B. Howe, and J. Heer. Formalizing visualization design knowledge as constraints: Actionable and extensible models in draco. IEEE Trans. Vis. Comp. Graph., 25(1):438–448, 2018

    2018

  19. [28]

    S. Pinker. A theory of graph comprehension. Artificial Intelligence and the Future of Testing, pp. 73–126, 1990

    1990

  20. [29]

    Poco and J

    J. Poco and J. Heer. Reverse-engineering visualizations: Recover- ing visual encodings from chart images. Comput. Graph. Forum., 36(3):353–363, 2017. doi: 10.1111/cgf.13193

    2017 doi

  21. [30]

    Y . Sun, J. Leigh, A. Johnson, and S. Lee. Articulate: A semi-automated model for translating natural language queries into meaningful visual- izations. In Proc. Int. Symp. Smart Graph., pp. 184–195, 2010. doi: 10. 1007/978-3-642-13544-6 18

    2010

  22. [31]

    J. Wei, X. Wang, et al. Chain-of-thought prompting elicits reasoning in large language models. Proc. NeurIPS, 35:24824–24837, 2022. doi: 10.5555/3600270.3602070

    2022

  23. [32]

    J. Wei, N. Xu, G. Chang, Y . Luo, B. Yu, and R. Guo. mChartQA: A universal benchmark for multimodal chart question answer based on vision-language alignment and reasoning, 2024. doi: 10.48550/arXiv. 2404.01548

    2024 doi arXiv

  24. [33]

    Wongsuphasawat, D

    K. Wongsuphasawat, D. Moritz, A. Anand, J. D. Mackinlay, B. Howe, and J. Heer. V oyager: Exploratory analysis via faceted browsing of visualization recommendations. IEEE Trans. Vis. Comp. Graph., 22(1):649–658, 2016. doi: 10.1109/TVCG.2015.2467191

    2016

  25. [34]

    R. Xia, B. Zhang, et al. ChartX & ChartVLM: A versatile benchmark and foundation model for complicated chart reasoning, 2024. doi: 10. 48550/arXiv.2402.12185

    2024 doi arXiv

  26. [35]

    Yu and C

    B. Yu and C. T. Silva. FlowSense: A natural language interface for visual data exploration within a dataflow system.IEEE Trans. Vis. Com- put. Graph., 26(1):1–11, 2020. doi: 10.1109/TVCG.2019.2934668

    2020

  27. [36]

    Yujian and L

    L. Yujian and L. Bo. A normalized levenshtein distance metric. IEEE Trans. Pattern Anal. Mach. Intell., 29(6):1091–1095, 2007

    2007

  28. [37]

    Zhang, A

    L. Zhang, A. Hu, et al. TinyChart: Efficient chart understanding with visual token merging and program-of-thoughts learning, 2024. doi: 10. 48550/arXiv.2404.16635

    2024 doi arXiv

  29. [38]

    Zhang, B

    Y . Zhang, B. Coecke, and M. Chen. MI3: Machine-initiated intelligent interaction for interactive classification and data reconstruction. ACM Transactions on Interactive Intelligent Systems, 11(3–4), Aug. 2021

    2021

  30. [39]

    Zhang, R

    Y . Zhang, R. Jiang, L. Xie, Y . Zhao, C. Liu, T. Ding, S. Chen, and X. Yuan. OldVisOnline: Curating a dataset of historical visualizations. IEEE Trans. Vis. Comput. Graph., 30(1):551–561, 2024. doi: 10.1109/ TVCG.2023.3326908

    2024

  31. [2019]

    doi: 10.1109/TVCG.2019.2934431

    2019

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.