Pith. sign in

REVIEW 4 major objections 5 minor 69 references

Urbanite: A Dataflow-Based Framework for Human-AI Interactive Alignment in Urban Visual Analytics

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Urbanite claims that non-programming urban experts can author visual analytics workflows by stating intent in natural language, provided the LLM's output is materialized as an editable, versioned dataflow.

desk verdict Urbanite is a thoughtful integration of LLMs with dataflow authoring for urban VA, but the 'executable workflows' claim is not backed by execution tests; the paper deserves a serious referee but needs a baseline and runtime evidence. read the letter →

arxiv 2508.07390 v1 pith:EOQSF4RV submitted 2025-08-10 cs.HC cs.AI

Gustavo Moreira , Leonardo Ferreira , Carolina Veiga , Maryam Hosseini , Fabio Miranda This is my paper · ORCID
classification cs.HCcs.AI
keywords urbanvisualanalyticsdataflowlargelanguagemodelshuman-AIalignmentnaturalinterfaceprovenanceintentspecificationvisualizationauthoring
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

Urbanite is a framework for authoring urban visual analytics systems by expressing analytical goals in natural language rather than writing code. The paper's central claim is that a dataflow—a graph of data-processing, modeling, and visualization components—can serve as an interpretable intermediary that keeps the user's high-level intent aligned with the code an LLM generates to realize it. Users specify intent at three granularities (the entire dataflow, a single node, or one parameter), and the framework keeps these levels synchronized, while provenance snapshots let users inspect, branch, and revert. The paper argues this lowers the barrier for domain experts to build, adapt, and share their own visual analytics workflows, and supports the claim with design goals drawn from a survey of system builders, three usage scenarios co-developed with urban experts, and a quantitative comparison against ten published urban analytics systems. The paper itself reports three cases where the generated task misread the original intent—inferring unavailable datasets or omitting a simulation step—and lists remaining limitations in automatic dataflow layout and visualization expressiveness.

What carries the argument

The central mechanism is the multi-scope dataflow specification: a machine-readable graph of nodes (data loading, transformation, modeling, visualization) and edges that carries, at three granularities, the user's intent—the dataflow-level task, each node's subtask and logic, and parameter-level annotations. This specification acts as the prompt context and output constraint for the LLM, as the object that UI, code, grammar, and natural-language edits all synchronize onto, and as the unit of versioning in the provenance history. It carries the argument because it turns an opaque LLM suggestion into an inspectable, editable, revertable artifact, making alignment a structured dialogue instead

What would settle it

Run Urbanite's generation pipeline repeatedly on the ten intents from the evaluation and count how many generated dataflows execute without human repair. A finding that a majority need structural edits—missing nodes, wrong edges, or unexecutable code—rather than parameter tweaks would falsify the claim that high-level intent translates into executable workflows.

Watch

Extended reading notes

Core claim

The core claim is that intent-based, LLM-assisted authoring of urban visual analytics can be made reliable by making the dataflow itself the shared artifact of human-AI alignment. Instead of treating the LLM as a black-box generator of final code, Urbanite has the LLM produce a structured dataflow specification—task, subtasks, nodes, edges, and parameter annotations—that the user can read, edit, and accept piece by piece. The multi-scope specification (dataflow level, node level, parameter level) is continuously synchronized so that editing a subtask updates the high-level task, and editing code updates the subtask description. Provenance snapshots record every LLM-accepted or user-made chan

Load-bearing premise

The framework assumes that the LLM produces structurally coherent, executable dataflow specifications from the user's intent; if the generated dataflows are flawed in structure, the entire multi-scope alignment mechanism—which treats the dataflow as the shared interpretation—fails.

Editorial extensions

If this is right

  • Urban researchers without programming expertise can author bespoke visual analytics workflows from natural-language goals, shifting them from end-users of systems to active builders.
  • Because every LLM suggestion is materialized as dataflow nodes and edges, users can validate the AI's interpretation before code runs, making errors visible and correctable rather than silent.
  • Task–subtask synchronization means changes at high-level goals propagate down to node logic, and node-level edits propagate back up, keeping documentation aligned with code.
  • Provenance snapshots create a branchable history of the analysis, supporting comparison of alternative scenarios, reversion, and reproducible sharing with collaborators.
  • Imperfect one-shot generation is acceptable by design: the framework's value lies in progressive refinement, not in flawless first outputs.

Reading between the lines

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

  • The same dataflow-as-mediator pattern should transfer to other domain-specific visual analytics fields with a stable component taxonomy, such as biological or geospatial analysis, where an LLM could generate drafts within a checkable graph.
  • A testable extension is to add automated structural checks—input/output type validation and trial execution—before a generated dataflow is presented to the user; the three reported mismatches suggest such checks could catch missing steps (e.g., simulation control) automatically.
  • Error rates likely vary by operation type: domain-specific and less common operations such as running a flood simulation may be more prone to omission or hallucination than standard data wrangling, a hypothesis one could measure across many intents.
  • If the framework is model-agnostic, the same prompts and schema could run on smaller or open-weight LLMs, which would matter for institutions with data-privacy constraints—though the paper does not test this.
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 / 5 minor

Summary. Urbanite is a dataflow-based framework for authoring urban visual analytics (VA) systems through LLM-mediated natural language interaction. It builds on the authors' prior Curio/UTK toolkits and adds multi-scope intent specification (dataflow, module, parameter levels), a structured specification layer, provenance, explanations, and LLM-driven task/subtask/code generation, operationalized around Terry et al.'s specification–process–evaluation alignment stages. The paper reports three usage scenarios co-developed with urban experts, feedback from six expert interviews, and a quantitative evaluation in which five evaluators scored LLM-generated task/subtask/dataflow outputs for ten external VA papers on semantic alignment, coverage, and flow quality. The authors conclude that Urbanite can 'effectively translate high-level user intent into executable workflows.'

Significance. If the central claim is validated, the paper makes a useful contribution to lowering programming barriers for urban VA authoring: it combines a dataflow intermediate representation with LLM assistance, supports iteration across abstraction levels, and is released as open source. The use of ten external papers' intents for quantitative scoring is a reasonable attempt at a corpus-based evaluation, and the explicit alignment/stage framing is a strength. However, the evidence currently does not establish that the generated workflows are executable, and the quantitative evaluation lacks a baseline and inter-rater reliability. The framework and the reported expert feedback are promising, but the headline 'executable workflows' claim is ahead of the data.

major comments (4)
  1. [§6.3 and §7] The central claim 'effectively translate high-level user intent into executable workflows' is not supported by the evaluation. Flow quality in §6.3 is a human judgment of structural coherence on a 0–2 scale, not a runtime check; there are no execution logs, sandbox runs, or error reports for the nine flows besides the explicitly flagged Konev et al. case, which 'lacked sufficient detail to support execution.' The conclusion should either be tested by running each generated flow and reporting success/failure/error rates, or softened to 'structurally plausible starting points requiring refinement.'
  2. [§6.3] The quantitative evaluation has no baseline and no inter-rater reliability. Five evaluators each score a subset of ten papers on an ordinal 0–2 scale; the paper reports only means and SDs (alignment 1.65, coverage 1.6, flow 1.5). Without a baseline such as direct LLM generation without Urbanite's specification layer, or a comparison to human-authored dataflows, these numbers do not establish that Urbanite is effective. Cohen's kappa or a similar agreement measure is needed, and per-paper scores should be reported.
  3. [§6.1 and §6.2] The usage scenarios are 'co-developed with urban experts who co-authored this paper,' and the expert feedback study uses only one scenario with six participants. This is an illustrative demonstration rather than an independent evaluation. The paper should label it as such and avoid generalizing beyond 'experts found the approach promising in a single walkthrough.' Independent scenario authorship and at least a second scenario would substantially increase confidence.
  4. [§5.4.1 and §6.3] The task/subtask synchronization and the generated dataflow are the load-bearing components of the alignment mechanism, but their correctness is not directly measured. The evaluation scores natural-language descriptions, not whether the synchronized task/subtask update actually preserves coherence across code, grammar, and dataflow. A concrete test would be to introduce a user edit and measure whether downstream generated subtasks/code remain consistent; this is currently untested.
minor comments (5)
  1. [Figure 1 caption] Typo: 'T ransform and analyze' should be 'Transform and analyze'; also 'throughprovenance' is missing a space.
  2. [§6.3] Use consistent notation for standard deviation (e.g., 'SD = 0.32') and provide the per-paper score table in the main text or supplementary; the current high-level aggregates are hard to interpret.
  3. [§5.5] The paper repeatedly references 'supplementary material' for the full specification and key differences from Curio/UTK. The submitted version should include this material or clearly state where it can be accessed, since the specification is central to reproducibility.
  4. [Table 1] Table 1 lists scope/mode combinations but gives no examples. Adding one concrete example per cell (e.g., 'NL: “map median income to color”') would make the interaction space much easier to understand.
  5. [§1] The phrase 'we demonstrate how such integration lowers the barrier... to translate high-level goals into executable workflows' also appears in the introduction. If the evaluation is revised to address executability, update this wording consistently.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the central evaluation is grounded in external papers and independent evaluators; only minor self-referentiality in usage scenarios.

full rationale

Urbanite's load-bearing claim—that LLM-generated dataflows align with user intent—is evaluated against core intents manually extracted from ten external urban VA papers, scored by five evaluators who are not paper authors (Sec. 6.3). This is an external benchmark, not a fitted input or a renamed output. The LLM pipeline (gpt-4o-mini with schema prompting) has no parameters fitted to the evaluation data. Self-citations to Curio [36] and UTK [35] are normal foundations: they describe the dataflow model and visualization grammar, but the effectiveness claim does not reduce to those citations. The only self-referential element is that the usage scenarios (Sec. 6.1) were co-developed with urban experts who are co-authors; this weakens qualitative evidence but is not a definitional reduction. The 'flow quality' metric measures human judgment of structural coherence rather than actual sandbox execution, so the conclusion's word 'executable' is broader than the evidence; that is a validity/evidence gap, not circularity. No equation, fitted parameter, or self-citation chain forces the paper's conclusion.

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

No numerical parameters are fit to data in this systems paper. The design choices (LLM model gpt-4o-mini, prompt templates, schema checks, 3-point rubric) are implementation decisions, not fitted scientific constants. No new physical or mathematical entities are postulated; 'scopes of specification', 'modes of expression', and the specification language are software/UX constructs.

assumptions (4)
  • domain assumption The survey of 10 visualization experts is representative of challenges in urban VA system building.
    Design goals DG1-DG6 are derived from this survey (Section 4); if the sample is unrepresentative, the framework's feature priorities lack empirical grounding.
  • domain assumption The interactive alignment framework of Terry et al. [52] (specification/process/evaluation) adequately captures human-AI alignment in dataflow authoring.
    Urbanite's architecture operationalizes this framework (Section 5.1); if the framework is wrong, the design goals do not target the right failure modes.
  • domain assumption Manually extracted user intents from 10 prior papers are faithful ground truth for the quantitative evaluation.
    Section 6.3 relies on these extractions; biased or inaccurate extraction would invalidate the alignment scores.
  • domain assumption The Curio dataflow model is a suitable formal substrate for LLM-assisted authoring.
    Urbanite adopts Curio's dataflow formalism (Section 5.3.1); if dataflows are not the right intermediate representation, the whole framework is undermined.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Urbanite: A Dataflow-Based Framework for Human-AI Interactive Alignment in Urban Visual Analytics." pith.science (2026). https://pith.science/paper/EOQSF4RV

@misc{pith2026250807390,
  author       = {Pith},
  title        = {Pith review of: Urbanite: A Dataflow-Based Framework for Human-AI Interactive Alignment in Urban Visual Analytics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EOQSF4RV}},
  note         = {Machine review of arXiv:2508.07390}
}
read the original abstract

With the growing availability of urban data and the increasing complexity of societal challenges, visual analytics has become essential for deriving insights into pressing real-world problems. However, analyzing such data is inherently complex and iterative, requiring expertise across multiple domains. The need to manage diverse datasets, distill intricate workflows, and integrate various analytical methods presents a high barrier to entry, especially for researchers and urban experts who lack proficiency in data management, machine learning, and visualization. Advancements in large language models offer a promising solution to lower the barriers to the construction of analytics systems by enabling users to specify intent rather than define precise computational operations. However, this shift from explicit operations to intent-based interaction introduces challenges in ensuring alignment throughout the design and development process. Without proper mechanisms, gaps can emerge between user intent, system behavior, and analytical outcomes. To address these challenges, we propose Urbanite, a framework for human-AI collaboration in urban visual analytics. Urbanite leverages a dataflow-based model that allows users to specify intent at multiple scopes, enabling interactive alignment across the specification, process, and evaluation stages of urban analytics. Based on findings from a survey to uncover challenges, Urbanite incorporates features to facilitate explainability, multi-resolution definition of tasks across dataflows, nodes, and parameters, while supporting the provenance of interactions. We demonstrate Urbanite's effectiveness through usage scenarios created in collaboration with urban experts. Urbanite is available at https://urbantk.org/urbanite.

Figures

Figures reproduced from arXiv: 2508.07390 by the authors.

Figure 1
Figure 1. Using Urbanite to analyze flood simulations. (a) The user loads a building layer and examines outputs through [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Urbanite’s interface, with design features for human-AI alignment. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Using Urbanite to analyze Project Sidewalk accessibility data. (a) The user starts with [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Using Urbanite to reproduce Urban Pulse. (a) The user writes code to load and parse pulse data. (b) Background layers are added, and [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

69 extracted references · 62 canonical work pages

  1. [1]

    Akbaba, D

    D. Akbaba, D. Lange, M. Correll, A. Lex, and M. Meyer. Troubling col- laboration: Matters of care for visualization design study. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems, article no. 812, pp. 1–15, 2023. 1, 2

    work page 2023

  2. [2]

    Amershi, D

    S. Amershi, D. Weld, M. V orvoreanu, A. Fourney, B. Nushi, P. Collisson, J. Suh, S. Iqbal, P. N. Bennett, K. Inkpen, et al. Guidelines for human- AI interaction. InProceedings of the 2019 CHI Conference on Human Factors in Computing Systems, pp. 1–13, 2019. 3

    work page 2019

  3. [3]

    Andrienko, N

    G. Andrienko, N. Andrienko, W. Chen, R. Maciejewski, and Y . Zhao. Visual analytics of mobility and transportation: State of the art and further research directions.IEEE Transactions on Intelligent Transportation Systems, 18(8):2232–2249, 2017. 2

    work page 2017

  4. [4]

    R. C. Basole and T. Major. Generative AI for visualization: Opportunities and challenges.IEEE Computer Graphics and Applications, 44(2):55–64,

    work page

  5. [5]

    Boorboor, Y

    S. Boorboor, Y . Kim, P. Hu, J. M. Moses, B. A. Colle, and A. E. Kaufman. Submerse: Visualizing storm surge flooding simulations in immersive display ecologies.IEEE Transactions on Visualization and Computer Graphics, 30(9):6365–6377, 2024. 8

    work page 2024

  6. [6]

    Büschel, A

    W. Büschel, A. Lehmann, and R. Dachselt. MIRIA: A mixed reality toolkit for the in-situ visualization and analysis of spatio-temporal interaction data. InProceedings of the 2021 CHI Conference on Human Factors in Computing Systems, article no. 470, pp. 1–15, 2021. 3

    work page 2021

  7. [7]

    S. P. Callahan, J. Freire, E. Santos, C. E. Scheidegger, C. T. Silva, and H. T. V o. VisTrails: Visualization meets data management. InProceedings of the 2006 ACM SIGMOD International Conference on Management of Data, pp. 745–747, 2006. 3

    work page 2006

  8. [8]

    J. Chen, Q. Huang, C. Wang, and C. Li. SenseMap: Urban performance visualization and analytics via semantic textual similarity.IEEE Transac- tions on Visualization and Computer Graphics, 30(9):6275–6290, 2024. 3

    work page 2024

Show all 69 references
  1. [9]

    M. Chen, J. Tworek, H. Jun, Q. Yuan, H. P. D. O. Pinto, J. Kaplan, H. Edwards, Y . Burda, N. Joseph, G. Brockman, et al. Evaluating large language models trained on code.arXiv preprint arXiv:2107.03374, 2021. 2

    2021 arXiv

  2. [10]

    C. V . F. de Souza, S. M. Bonnet, D. de Oliveira, M. Cataldi, F. Miranda, and M. Lage. ProWis: A visual approach for building, managing, and analyzing weather simulation ensembles at runtime.IEEE Transactions on Visualization and Computer Graphics, 30(1):738–747, 2024. 8

    2024

  3. [11]

    Z. Deng, D. Weng, J. Chen, R. Liu, Z. Wang, J. Bao, Y . Zheng, and Y . Wu. AirVis: Visual analytics of air pollution propagation.IEEE Transactions on Visualization and Computer Graphics, 26(1):800–810, 2019. 3

    2019

  4. [12]

    Z. Deng, D. Weng, S. Liu, Y . Tian, M. Xu, and Y . Wu. A survey of urban visual analytics: Advances and future directions.Computational Visual Media, 9(1):3–39, 2023. 1, 2

    2023

  5. [13]

    J. Dong, H. Zhang, M. Cui, Y . Lin, H.-Y . Wu, and C. Bi. TCEVis: Visual analytics of traffic congestion influencing factors based on explainable machine learning.Visual Informatics, 8(1):56–66, 2024. 3

    2024

  6. [14]

    Elmqvist, J

    N. Elmqvist, J. Stasko, and P. Tsigas. DataMeadow: A visual canvas for analysis of large-scale multivariate data. In2007 IEEE Symposium on Visual Analytics Science and Technology, pp. 187–194. IEEE, 2007. 3

    2007

  7. [15]

    Z. Feng, H. Qu, S.-H. Yang, Y . Ding, and J. Song. A survey of visual analytics in urban area.Expert Systems, 39(9):e13065, 2022. 1, 2

    2022

  8. [16]

    Ferreira, G

    L. Ferreira, G. Moreira, M. Hosseini, M. Lage, N. Ferreira, and F. Miranda. Assessing the landscape of toolkits, frameworks, and authoring tools for urban visual analytics systems.Computers & Graphics, 123:104013, 2024. 1, 2, 3

    2024

  9. [17]

    Ferreira, M

    N. Ferreira, M. Lage, H. Doraiswamy, H. V o, L. Wilson, H. Werner, M. Park, and C. Silva. Urbane: A 3D framework to support data driven decision making in urban development. In2015 IEEE Conference on Visual Analytics Science and Technology (VAST), pp. 97–104. IEEE, 2015. 9

    2015

  10. [18]

    Ferreira, J

    N. Ferreira, J. Poco, H. T. V o, J. Freire, and C. T. Silva. Visual exploration of big spatio-temporal urban data: A study of New York City taxi trips. IEEE Transactions on Visualization and Computer Graphics, 19(12):2149– 2158, 2013. 9

    2013

  11. [19]

    Garcia, J

    G. Garcia, J. Silveira, J. Poco, A. Paiva, M. B. Nery, C. T. Silva, S. Adorno, and L. G. Nonato. CrimAnalyzer: Understanding crime patterns in São Paulo.IEEE Transactions on Visualization and Computer Graphics, 27(4):2313–2328, 2019. 1, 3

    2019

  12. [20]

    Javed and N

    W. Javed and N. Elmqvist. ExPlates: Spatializing interactive analysis to scaffold visual exploration.Computer Graphics Forum, 32(3pt4):441–450,

  13. [21]

    Jänicke, P

    S. Jänicke, P. Kaur, P. Kuzmicki, and J. Schmidt. Participatory visu- alization design as an approach to minimize the gap between research and application. In C. Gillmann, M. Krone, G. Reina, and T. Wischgoll, eds.,VisGap - The Gap between Visualization Research and Visualizat...

    2020

  14. [22]

    Konev, J

    A. Konev, J. Waser, B. Sadransky, D. Cornel, R. A. Perdigão, Z. Horváth, and M. E. Gröller. Run Watchers: Automatic simulation-based decision support in flood management.IEEE Transactions on Visualization and Computer Graphics, 20(12):1873–1882, 2014. 9

    2014

  15. [23]

    S. Lau, I. Drosos, J. M. Markel, and P. J. Guo. The design space of computational notebooks: An analysis of 60 systems in academia and industry. In2020 IEEE Symposium on Visual Languages and Human- Centric Computing (VL/HCC), pp. 1–11. IEEE, 2020. 2

    2020

  16. [24]

    J. Li, S. Chen, K. Zhang, G. Andrienko, and N. Andrienko. COPE: Interactive exploration of co-occurrence patterns in spatial time series. IEEE Transactions on Visualization and Computer Graphics, 25(8):2554– 2567, 2019. 3

    2019

  17. [25]

    S. L’Yi, A. van den Brandt, E. Adams, H. N. Nguyen, and N. Gehlen- borg. Learnable and expressive visualization authoring through blended interfaces.IEEE Transactions on Visualization and Computer Graphics, 31(1):459–469, 2025. 2, 3

    2025

  18. [26]

    S. L’Yi, Q. Wang, F. Lekschas, and N. Gehlenborg. Gosling: A grammar- based toolkit for scalable and interactive genomics data visualization.IEEE Transactions on Visualization and Computer Graphics, 28(1):140–150,

  19. [27]

    Y . Lyu, H. Lu, M. K. Lee, G. Schmitt, and B. Y . Lim. IF-City: Intelligible fair city planning to measure, explain and mitigate inequality.IEEE Transactions on Visualization and Computer Graphics, 30(7):3749–3766,

  20. [28]

    McKenna, D

    S. McKenna, D. Mazur, J. Agutter, and M. Meyer. Design activity frame- work for visualization design.IEEE Transactions on Visualization and Computer Graphics, 20(12):2191–2200, 2014. 3

    2014

  21. [29]

    A. M. McNutt. No grammar to rule them all: A survey of JSON-style DSLs for visualization.IEEE Transactions on Visualization and Computer Graphics, 29(1):160–170, 2023. 2

    2023

  22. [30]

    Minaee, T

    S. Minaee, T. Mikolov, N. Nikzad, M. Chenaghlu, R. Socher, X. Am- atriain, and J. Gao. Large language models: A survey.arXiv preprint arXiv:2402.06196, 2025. 2

    2025 arXiv

  23. [31]

    Miranda, H

    F. Miranda, H. Doraiswamy, M. Lage, L. Wilson, M. Hsieh, and C. T. Silva. Shadow Accrual Maps: Efficient accumulation of city-scale shadows over time.IEEE Transactions on Visualization and Computer Graphics, 25(3):1559–1574, 2018. 1

    2018

  24. [32]

    Miranda, H

    F. Miranda, H. Doraiswamy, M. Lage, K. Zhao, B. Gonçalves, L. Wilson, M. Hsieh, and C. T. Silva. Urban Pulse: Capturing the rhythm of cities. IEEE Transactions on Visualization and Computer Graphics, 23(1):791– 800, 2016. 3, 8

    2016

  25. [33]

    Miranda, M

    F. Miranda, M. Hosseini, M. Lage, H. Doraiswamy, G. Dove, and C. T. Silva. Urban Mosaic: Visual exploration of streetscapes using large-scale image data. InProceedings of the 2020 CHI Conference on Human Factors in Computing Systems, pp. 1–15, 2020. 3

    2020

  26. [34]

    Miranda, T

    F. Miranda, T. Ortner, G. Moreira, M. Hosseini, M. Vuckovic, F. Biljecki, C. T. Silva, M. Lage, and N. Ferreira. The state of the art in visual analytics for 3D urban data.Computer Graphics Forum, 43(3):e15112, 2024. 1, 2, 3

    2024

  27. [35]

    Moreira, M

    G. Moreira, M. Hosseini, M. N. A. Nipu, M. Lage, N. Ferreira, and F. Mi- randa. The Urban Toolkit: A grammar-based framework for urban visual analytics.IEEE Transactions on Visualization and Computer Graphics, 30(1):1402–1412, 2024. 2, 3, 4, 6

    2024

  28. [36]

    Moreira, M

    G. Moreira, M. Hosseini, C. Veiga, L. Alexandre, N. Colaninno, D. de Oliveira, N. Ferreira, M. Lage, and F. Miranda. Curio: A dataflow- based framework for collaborative urban visual analytics.IEEE Transac- tions on Visualization and Computer Graphics, 31(1):1224–1234, 2025. 2...

    2025

  29. [37]

    Munzner.Visualization analysis and design

    T. Munzner.Visualization analysis and design. CRC press, 2014. 2

    2014

  30. [38]

    J. Nielsen. AI: First new UI paradigm in 60 years. https://www. nngroup.com/articles/ai-paradigm/, 2023. Accessed: 2025-03-

    2023

  31. [39]

    R. Qiu, Y . Tu, Y .-S. Wang, P.-Y . Yen, and H.-W. Shen. DocFlow: A visual analytics system for question-based document retrieval and categorization. IEEE Transactions on Visualization and Computer Graphics, 30(2):1533– 1548, 2024. 3

    2024

  32. [40]

    S. Sah, R. Mitra, A. Narechania, A. Endert, J. Stasko, and W. Dou. Gener- ating analytic specifications for data visualization from natural language queries using large language models.arXiv preprint arXiv:2408.13391,

    arXiv

  33. [41]

    M. Saha, M. Saugstad, H. T. Maddali, A. Zeng, R. Holland, S. Bower, A. Dash, S. Chen, A. Li, K. Hara, et al. Project Sidewalk: A web-based crowdsourcing tool for collecting sidewalk accessibility data at scale. In Proceedings of the 2019 CHI Conference on Human Factors in Comp...

    2019

  34. [42]

    Salinas, T

    K. Salinas, T. Gonçalves, V . Barella, T. Vieira, and L. G. Nonato. CityHub: A library for urban data integration. In2022 35th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), vol. 1, pp. 43–48, 2022. 3

    2022

  35. [43]

    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,

  36. [44]

    Sicat, J

    R. Sicat, J. Li, J. Choi, M. Cordeil, W.-K. Jeong, B. Bach, and H. Pfister. DXR: A toolkit for building immersive data visualizations.IEEE Trans- actions on Visualization and Computer Graphics, 25(1):715–725, 2019. 2

    2019

  37. [45]

    C. T. Silva, E. Anderson, E. Santos, and J. Freire. Using VisTrails and provenance for teaching scientific visualization.Computer Graphics Forum, 30(1):75–84, 2011. 3

    2011

  38. [46]

    C. T. Silva, J. Freire, and S. P. Callahan. Provenance for visualizations: Re- producibility and beyond.Computing in Science & Engineering, 9(5):82– 89, 2007. 2

    2007

  39. [47]

    Silva, D

    V . Silva, D. de Oliveira, P. Valduriez, and M. Mattoso. Dfanalyzer: Run- time dataflow analysis of scientific applications using provenance.Pro- ceedings of the VLDB Endowment, 11(12):2082–2085, 2018. 3

  40. [48]

    Spinner, U

    T. Spinner, U. Schlegel, H. Schäfer, and M. El-Assady. explAIner: A visual analytics framework for interactive and explainable machine learning. IEEE Transactions on Visualization and Computer Graphics, 26(1):1064– 1074, 2020. 2

    2020

  41. [49]

    Steyvers, H

    M. Steyvers, H. Tejeda, A. Kumar, C. Belem, S. Karny, X. Hu, L. W. Mayer, and P. Smyth. What large language models know and what people think they know.Nature Machine Intelligence, pp. 1–11, 2025. 2

    2025

  42. [50]

    Z. Sun, D. Xu, Y . Zhang, Y . Qi, Y . Wang, Z. Zuo, Z. Wang, Y . Li, X. Li, Q. Lu, W. Peng, and S. Guo. BigDataflow: A distributed interproce- dural dataflow analysis framework. InProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the F...

    2023

  43. [51]

    X. Tang, E. Krokos, C. Liu, K. Davidson, K. Whitley, N. Ramakrishnan, and C. North. Steering LLM summarization with visual workspaces for sensemaking.arXiv preprint arXiv:2409.17289, 2024. 3

    2024 arXiv

  44. [52]

    Terry, C

    M. Terry, C. Kulkarni, M. Wattenberg, L. Dixon, and M. R. Morris. In- teractive AI alignment: Specification, process, and evaluation alignment. arXiv preprint arXiv:2311.00710, 2023. 2, 3, 4

    2023 arXiv

  45. [53]

    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,

  46. [54]

    Ulbrich, M

    P. Ulbrich, M. Waldner, K. Furmanová, S. M. Marques, D. Bednáˇr, B. Ko- zlíková, and J. Byška. sMolBoxes: Dataflow model for molecular dy- namics exploration.IEEE Transactions on Visualization and Computer Graphics, 29(1):581–590, 2022. 3

    2022

  47. [55]

    Walker, A

    R. Walker, A. Slingsby, J. Dykes, K. Xu, J. Wood, P. H. Nguyen, D. Stephens, B. W. Wong, and Y . Zheng. An extensible framework for provenance in human terrain visual analytics.IEEE Transactions on Visualization and Computer Graphics, 19(12):2139–2148, 2013. 2

    2013

  48. [56]

    S. Wang, F. Lyu, S. Wang, C. E. Catlett, A. Padmanabhan, and K. Soltani. Integrating CyberGIS and urban sensing for reproducible streaming ana- lytics.Urban Informatics, pp. 663–681, 2021. 3

    2021

  49. [57]

    Waser, H

    J. Waser, H. Ribicic, R. Fuchs, C. Hirsch, B. Schindler, G. Bloschl, and E. Groller. Nodes on Ropes: A comprehensive data and control flow for steering ensemble simulations.IEEE Transactions on Visualization and Computer Graphics, 17(12):1872–1881, 2011. 3

    2011

  50. [58]

    A. Wu, D. Deng, M. Chen, S. Liu, D. Keim, R. Maciejewski, S. Miksch, H. Strobelt, F. Viégas, and M. Wattenberg. Grand challenges in visual ana- lytics applications.IEEE Computer Graphics and Applications, 43(5):83– 90, 2023. 1, 2, 3

    2023

  51. [59]

    A. Wu, D. Deng, F. Cheng, Y . Wu, S. Liu, and H. Qu. In defence of visual analytics systems: Replies to critics.IEEE Transactions on Visualization and Computer Graphics, 29(1):1026–1036, 2023. 1

    2023

  52. [60]

    Xia and X

    X. Xia and X. Ming. SynxFlow: Synergising high-performance hazard simulation with data flow, 2025. 8

    2025

  53. [61]

    T. Xie, Y . Ma, J. Kang, H. Tong, and R. Maciejewski. FairRankVis: A visual analytics framework for exploring algorithmic fairness in graph min- ing models.IEEE Transactions on Visualization and Computer Graphics, 28(1):368–377, 2022. 3

    2022

  54. [62]

    Q. Yang, A. Steinfeld, C. Rosé, and J. Zimmerman. Re-examining whether, why, and how human-AI interaction is uniquely difficult to design. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, pp. 1–13, 2020. 3

    2020

  55. [63]

    W. Yap, P. Janssen, and F. Biljecki. Free and open source urbanism: Software for urban planning practice.Computers, Environment and Urban Systems, 96:101825, 2022. 3

    2022

  56. [64]

    Yu and C

    B. Yu and C. T. Silva. VisFlow-web-based visualization framework for tabular data with a subset flow model.IEEE Transactions on Visualization and Computer Graphics, 23(1):251–260, 2016. 3

    2016

  57. [65]

    Yu and C

    B. Yu and C. T. Silva. FlowSense: A natural language interface for visual data exploration within a dataflow system.IEEE Transactions on Visualization and Computer Graphics, 26(1):1–11, 2020. 3

    2020

  58. [66]

    Y . Zhao, Y . Zhang, Y . Zhang, X. Zhao, J. Wang, Z. Shao, C. Turkay, and S. Chen. LEV A: Using large language models to enhance visual analytics. IEEE Transactions on Visualization and Computer Graphics, 2024. 2, 3

    2024

  59. [67]

    Zheng, L

    Y . Zheng, L. Capra, O. Wolfson, and H. Yang. Urban computing: Concepts, methodologies, and applications.ACM Transactions on Intelligent Systems and Technology, 5(3):1–55, 2014. 2

    2014

  60. [68]

    Zheng, W

    Y . Zheng, W. Wu, Y . Chen, H. Qu, and L. M. Ni. Visual analytics in urban computing: An overview.IEEE Transactions on Big Data, 2(3):276–296,

  61. [69]

    Ziegler and S

    P. Ziegler and S. E. Chasins. A need-finding study with users of geospatial data. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems, article no. 862, pp. 1–16, 2023. 3

    2023

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.