Pith. sign in

REVIEW 3 major objections 5 minor 69 references

ManiScope: LLM-Assisted Visual Analytics of Cryptocurrency Manipulation Risk

T0 review · 3 major / 5 minor · reviewed 2026-07-14 · grok-4.5

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

Pith's one-line read ManiScope turns an LLM into a co-analyst that reads your crypto exploration and returns evidence for the hypotheses you are forming.

desk verdict Solid cs.HC systems paper: multi-view crypto manipulation VA plus a real co-analyst LLM loop; the user-study gains are real but only partly isolated from unequal automation. read the letter →

arxiv 2607.11451 v1 pith:LNTREQFZ submitted 2026-07-13 cs.HC

Xiaolin Wen , Feng Liang , Yuanye Ma , Qishuang Fu , Zhengyu Sun , Feng Zhu , Can Liu , Yong Wang This is my paper · ORCID
classification cs.HC
keywords cryptocurrencymanipulationvisualanalyticshuman-LLMcollaborationwashtradingon-chainanalysishypothesis-drivenholderrelationships
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

Trade-based crypto manipulation is hard to judge because strategies change, ground truth is scarce, and risk depends on combining holder structure, suspicious trades, and price moves. Fixed detectors and single-signal dashboards leave investors doing repetitive cross-view work. ManiScope is a visual analytics system with coordinated views of token holdings, multi-level holder relationships, rule-flagged wash and coordinated trades, price dynamics, and detailed holder behavior. On top of that interface, a human–LLM framework treats the model as a co-analyst: it infers intent and emerging hypotheses from interaction traces, annotations, and screenshots, plans follow-up visual and statistical checks, and returns findings that support, refine, or contradict those hypotheses. Case studies and a 12-person study with experienced practitioners suggest this reduces manual evidence hunting and organizes analysis around the user’s own hypotheses better than a reactive chat assistant on the same views.

What carries the argument

Human–LLM collaborative visual analytics framework: a bottom-up hypothesis inference stage from interaction traces and screenshots, a top-down analysis-planning stage that expands missing goals and tasks, and a finding-generation stage that executes visual and statistical actions and organizes supporting, refining, or contradicting evidence around user and derived hypotheses.

What would settle it

A controlled comparison in which the co-analyst framework, given real interaction traces, produces hypotheses and findings that experts rate as misaligned or irrelevant at high rates, or that fail to improve evidence organization and confidence versus a reactive assistant on the same interface and data.

Watch

Extended reading notes

Core claim

The paper claims that positioning an LLM as a co-analyst—inferring analytical intent and hypotheses from interaction context and returning visual, statistical, and synthesized evidence for hypothesis evaluation—supports flexible manipulation-risk assessment while reducing repetitive inspection and strengthening evidence-based reasoning, compared with a reactive LLM assistant that shares the same visual analytics substrate.

Load-bearing premise

The system assumes that clicks, notes, and screenshots are a faithful enough window into what the analyst is really trying to test that auto-inferred hypotheses and follow-up findings stay mostly on-target rather than noisy or leading.

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 / 5 minor

Summary. ManiScope is an LLM-assisted visual analytics system for trade-based cryptocurrency manipulation risk. It combines a configurable multi-view VA interface (token distribution/relationships, price dynamics with rule-flagged wash and same-direction patterns, holder behavior timelines, and user-defined entity/relationship/flagging rules) with a human–LLM co-analysis framework that infers hypotheses from interaction traces, annotations, and screenshots, plans follow-up visual/statistical checks, and returns supporting, refining, or contradicting findings. Design is grounded in interviews with four practitioners (R1–R6). Evaluation comprises two expert case studies on ACT and PNUT and a counterbalanced within-subject study with 12 practitioners against a same-interface reactive LLM baseline, reporting higher ratings on evidence finding/organization/confidence, high checklist rates for hypothesis alignment and finding relevance, and qualitative tradeoffs (structure vs. latency/irrelevant context).

Significance. If the comparative claim holds, the paper contributes a domain-grounded multi-view VA system for manipulation-risk assessment and a concrete co-analyst pattern that goes beyond reactive chat by tying LLM work to hypothesis-driven knowledge generation. Strengths include practitioner-derived requirements, explicit parameterized evidence modeling, a capability-rich agent design (Appendix B roles/tools), a same-interface baseline, and multi-method evaluation (cases, Likert, output checklist, interviews). The work is relevant to HCI/VA and FinTech visualization, and the framework generalization discussion is useful even if domain-specific rules do not transfer.

major comments (3)
  1. [§VII-A / Appendix B] §VII-A baseline specification is load-bearing for the central comparative claim (intent-inferring co-analyst > reactive LLM on the same VA interface). The baseline is described as a simplified reactive assistant that can access analysis state, answer questions, and execute user-specified requests, while ManiScope alone receives the full hypothesis-inference/planning/finding pipeline and multi-branch sub-agents with visual, statistical, and model-parameter tools (Appendix B). Without a capability-matched reactive control (same tools/scripts/parameter sweeps, only user-initiated), the significant gains on LLM2–LLM4 (p=.043/.012/.019) may reflect unequal agency and automation rather than faithful intent mapping from traces/screenshots. Please specify baseline tool surface, prompt budget, and executable actions, or add/report a matched-capability ablation.
  2. [§V-B / Table II / §VII-B] §V-B and R5 rest on the assumption that interactions, annotations, and screenshots sufficiently encode analytical intent. Table II and MS1/MS3 show mostly positive but imperfect context capture; qualitative feedback reports irrelevant context and latency. The paper should quantify failure modes of hypothesis inference (false/leading hypotheses, unsupported expansions) and report how often users rejected or ignored LLM-derived branches, not only aggregate alignment/relevance rates. This is needed to support that the co-analyst strengthens risk assessment rather than mainly increasing automated exploration volume.
  3. [§VI–§VII] Evaluation does not establish external validity of risk conclusions. There is no ground-truth or expert-adjudicated manipulation labels for ACT/PNUT, no inter-rater agreement on risk assessments across conditions, and no measure of decision quality beyond self-reported confidence and checklist relevance. Given the paper’s framing as risk assessment rather than detection, at least report consistency of conclusions across participants/conditions and whether ManiScope changed risk judgments relative to baseline, not only process metrics.
minor comments (5)
  1. [§VII-B1] Ease-of-learning is the weakest VA rating (VA6 M=5.25); a short onboarding/complexity analysis would help readers judge deployability of the multi-view + agent workflow.
  2. [§V-A2 / §VI-A] Free parameters (top-holder threshold, entity/link rules, Round Trip / Same Direction thresholds) are central to C1/R1–R3; a sensitivity summary beyond the ACT case parameter sweep would strengthen claims of flexible evidence modeling.
  3. [Appendix B] Appendix B is valuable but dense; a compact main-text table of agent roles, tools, and what the baseline could/could not invoke would clarify the experimental contrast.
  4. [Fig. 4–5] Figures 4–5 are informative but text-heavy; ensure readable encodings and explicit callouts for supporting vs. contradicting findings in the camera-ready version.
  5. [§II-C] Related work on LLM-assisted VA is current; a sharper one-paragraph contrast with ProactiveVA / LightVA / LEVA on hypothesis inference vs. proactive suggestion would help position novelty.

Circularity Check

1 steps flagged · score 1.0 of 10

No derivation circularity: claims rest on comparative user ratings and case narratives, not equations that redefine the target; only a mild evaluation tautology on alignment of trace-inferred hypotheses.

  1. other [§VII-B2 Table II; LLM output checklist OC1; §V-B Hypothesis Inference]
    "The checklist results were positive, with high rates for hypothesis alignment (96.2%)... This source breakdown shows... user-trace inferred hypotheses and findings (20 and 181)... the agent takes users' interaction sequences with screenshots of the current visual context as input and progressively infers hypotheses"

    For the 20/26 user-trace-inferred hypotheses, alignment with the participant's analysis context is scored against the same interaction traces/annotations/screenshots used as LLM input. High OC1 alignment for that subset is partly expected if the model restates the trace, so the metric is weakly self-referential. It is not load-bearing for the paper's comparative claim (ManiScope vs reactive baseline on LLM1–LLM5) and does not redefine risk assessment by construction.

full rationale

ManiScope is a systems/HCI paper, not a first-principles derivation. Its strongest claims (effective risk assessment, reduced manual evidence-seeking, better organization around hypotheses than a reactive LLM baseline) are supported by a within-subject user study (n=12), questionnaires (VA1–VA7, LLM1–LLM5, MS1–MS4), an LLM-output checklist, two case studies, and qualitative feedback—not by fitted parameters renamed as predictions or uniqueness theorems imported from the authors. Related-work self-citations (e.g., NFTDisk, PonziLens+) are background on crypto visualization and are not load-bearing for the co-analyst claim; the knowledge-generation model is external (Sacha et al.). The only mild circularity-adjacent pattern is that hypothesis-alignment rates (96.2%) partly score user-trace-inferred hypotheses against the same interaction context used to generate them, so high alignment for that subset is partly expected by construction; this does not force the comparative gains vs. baseline or the finding-sufficiency/relevance results. Baseline capability mismatch is a validity concern, not circularity. Score 1 reflects that minor evaluation tautology without elevating it to a forced central claim.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

Load-bearing content is mostly domain and design assumptions, not fitted physical constants. The scientific claim depends on treating heuristic on-chain patterns as risk evidence, treating interaction traces as intent, and treating practitioner Likert/checklist judgments as validation of ‘effective risk assessment.’ Free parameters are user/system thresholds that change what is flagged as suspicious; invented entities are the system and framework themselves.

free parameters (3)
  • top-holder cumulative balance threshold (e.g., 0.3) = example 0.3 (30% cumulative holdings)
    Chooses which addresses appear as major holders; case study and LLM sweeps show concentration findings are threshold-dependent.
  • entity/relationship rule parameters (network, similarity, suspicious-pattern)
    User-chosen strictness defines who is ‘same entity’ or ‘related’; directly shapes the Token Distribution and Behavior views that feed risk conclusions.
  • wash-trading Round Trip and Same Direction flagging thresholds
    Sequence length, time windows, near-zero net position/earnings tolerances determine which patterns are labeled suspicious and shown in the Manipulation View.
assumptions (5)
  • domain assumption Trade-based manipulation risk can be usefully assessed from on-chain DEX trades/transfers without definitive proof of intent.
    Stated in §III and framing throughout; evaluation treats suspicious patterns as risk evidence, not ground-truth labels.
  • domain assumption Wash trading and coordinated same-direction trading are the primary operational proxies for trade-based manipulation in this system.
    §III and R3; other schemes (e.g., pump-and-dump social coordination) are background only.
  • ad hoc to paper User interactions, annotations, and screenshots encode analytical intent well enough for bottom-up hypothesis inference (R5).
    Core of the co-analyst framework in §V-B; not independently validated outside this system’s checklist.
  • domain assumption The knowledge generation model for visual analytics (Sacha et al.) is an appropriate scaffold for placing LLM co-analysis in the loop.
    Cited and used as conceptual grounding in §V-B and Discussion.
  • domain assumption Practitioner subjective risk assessment and Likert/checklist ratings are valid primary success criteria when ground truth is scarce.
    §VII study design explicitly avoids narrowly scored detection tasks for this reason.
invented entities (2)
  • ManiScope base visual analytics system (Token Distribution, Manipulation, Behavior Detail, Control Panel)
    purpose: Flexible multi-view modeling of holders, relationships, suspicious patterns, and market dynamics for manipulation-risk analysis.
    New integrated interface contribution; independent_evidence false beyond this paper’s cases/study.
  • Human–LLM collaborative VA framework (hypothesis inference, analysis planning, finding generation with support/skeptical/expansion sub-agents)
    purpose: Position the LLM as co-analyst that reduces repetitive hypothesis–evidence iteration and structures evidence around user hypotheses.
    Central methodological invention; evaluated only inside ManiScope sessions, not as a portable formal theory.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ManiScope: LLM-Assisted Visual Analytics of Cryptocurrency Manipulation Risk." pith.science (2026). https://pith.science/paper/LNTREQFZ

@misc{pith2026260711451,
  author       = {Pith},
  title        = {Pith review of: ManiScope: LLM-Assisted Visual Analytics of Cryptocurrency Manipulation Risk},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LNTREQFZ}},
  note         = {Machine review of arXiv:2607.11451}
}
read the original abstract

Cryptocurrency markets are vulnerable to trade-based manipulation, such as wash trading, which can distort price signals and mislead investors. Prior research has mainly focused on detecting manipulation using fixed rules or labeled examples, offering limited flexibility and interpretability for assessing potential risks. Existing visual analytics tools can reveal basic manipulation-related signals, such as token distribution, but still require substantial manual effort to integrate holder relationships, suspicious behaviors, and market dynamics for risk assessment. To address these limitations, we propose ManiScope, an LLM-assisted visual analytics system for analyzing trade-based manipulation risks in cryptocurrency markets. ManiScope provides coordinated views of token distributions, holder relationships, detailed holder behaviors, price dynamics, and suspicious trading patterns. To further enhance user analysis, ManiScope introduces a human-LLM collaborative visual analytics framework. Rather than acting as a basic reactive LLM assistant, the framework positions the LLM as a co-analyst that infers users' analytical intent and emerging hypotheses from interaction context and surfaces relevant visual, statistical, and synthesized evidence for hypothesis evaluation. This design reduces repetitive inspection and strengthens evidence-based reasoning. We evaluate ManiScope through two case studies and a user study with 12 experienced cryptocurrency practitioners. The results suggest that ManiScope supports effective risk assessment of manipulation, reduces manual effort in evidence-seeking, and organizes findings around user hypotheses.

Figures

Figures reproduced from arXiv: 2607.11451 by the authors.

Figure 1
Figure 1. The ManiScope interface. (A) The base visual analytics system consists of the Token Distribution View (A1), Manipulation View (A2), Behavior Detail View (A3), and Control Panel (A4). (B) The human-LLM collaborative interface augments the base system with an LLM analysis panel (B1), user action and annotation histories (B2–B3), an action tree for tracking the analytical process (B4), and a chat panel for interactive … view at source ↗
Figure 2
Figure 2. Overview of ManiScope’s human-LLM collaborative visual analytics framework. The framework combines a shared data-model-visualization substrate, [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. The LLM workflow in ManiScope consists of three stages: bottom-up hypothesis inference and generation from interaction context, top-down analysis planning for expanding the analytical hierarchy, and finding generation that produces visual, statistical, and synthesized findings for hypothesis verification. For the context-aware LLM analysis workflow, the LLM is embedded into the exploration and verification process a… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: A case showing how LLM assistance reduced manual follow-up in ACT. P1 explored the token distribution (A), behavior details (B), and suspicious [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: A case showing how LLM assistance made PNUT analysis more evidence-based. P2 explored the suspicious-pattern timeline (A), expanded a selected [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Questionnaire results from the user study, grouped by the base visual [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

69 extracted references · 5 linked inside Pith

  1. [1]

    Market mis- conduct in decentralized finance (DeFi): Analysis, regulatory challenges and policy implications,

    X. Xiong, Z. Wang, T. Cui, W. Knottenbelt, and M. Huth, “Market mis- conduct in decentralized finance (DeFi): Analysis, regulatory challenges and policy implications,”arXiv preprint arXiv:2311.17715, 2023

    arXiv 2023

  2. [2]

    Cryptocurrency market manipulation–a systematic literature review,

    F. Eigelshoven, A. Ullrich, and D. Parry, “Cryptocurrency market manipulation–a systematic literature review,” inProceedings of the International Conference on Information Systems, 2021, p. 1

    2021

  3. [3]

    Market manipulation: Suspected wash trading and pump-and-dump schemes,

    Chainalysis Team, “Market manipulation: Suspected wash trading and pump-and-dump schemes,” https://www.chainalysis.com/blog/ crypto-market-manipulation-wash-trading-pump-and-dump-2025/, 2025, accessed: 2026-05-22

    2025

  4. [4]

    Wash trading at cryptocurrency exchanges,

    G. L. Pennec, I. Fiedler, and L. Ante, “Wash trading at cryptocurrency exchanges,”Finance Research Letters, vol. 43, p. 101982, 2021

    2021

  5. [5]

    A survey on pump and dump detection in cryptocurrency markets using machine learning,

    M. J. Rajaeiet al., “A survey on pump and dump detection in cryptocurrency markets using machine learning,”Future Internet, vol. 15, no. 8, p. 267, 2023. [Online]. Available: https://www.mdpi. com/1999-5903/15/8/267

    2023

  6. [6]

    The chainalysis 2025 crypto crime report,

    Chainalysis, “The chainalysis 2025 crypto crime report,” https: //go.chainalysis.com/2025-Crypto-Crime-Report.html, 2025, accessed: 2026-03-01

    2025

  7. [7]

    Crypto wash trading,

    L. W. Cong, X. Li, K. Tang, and Y . Yang, “Crypto wash trading,” Management Science, vol. 69, no. 11, pp. 6427–6454, 2023. 13

    2023

  8. [8]

    Determinants of wash trading in major cryptoexchanges,

    J. Sila, E. Ko ˇcenda, L. Kristoufek, and J. Kukacka, “Determinants of wash trading in major cryptoexchanges,”Available at SSRN 4971590, 2025

    2025

Show all 69 references
  1. [9]

    To the moon: defining and detecting cryptocurrency pump-and-dumps,

    J. Kamps and B. Kleinberg, “To the moon: defining and detecting cryptocurrency pump-and-dumps,”Crime Science, vol. 7, no. 1, pp. 1– 18, 2018

    2018

  2. [10]

    Detecting and quantifying wash trading on decentralized cryptocurrency exchanges,

    F. Victor and A. M. Weintraud, “Detecting and quantifying wash trading on decentralized cryptocurrency exchanges,” inProceedings of the Web Conference 2021, 2021, pp. 23–32

    2021

  3. [11]

    Wash trading and insider sales in nft markets,

    S. Wang, N. Cheng, and T. Zhang, “Wash trading and insider sales in nft markets,”Journal of Banking & Finance, p. 107529, 2025

    2025

  4. [12]

    Beyond the surface: advanced wash- trading detection in decentralized nft markets,

    A. To ˇsi´c, J. Viˇciˇc, and N. Hrovatin, “Beyond the surface: advanced wash- trading detection in decentralized nft markets,”Financial Innovation, vol. 11, no. 1, pp. 1–21, 2025

    2025

  5. [13]

    Detecting cryp- tocurrency pump-and-dump frauds using market and social signals,

    H. Nghiem, G. Muric, F. Morstatter, and E. Ferrara, “Detecting cryp- tocurrency pump-and-dump frauds using market and social signals,” Expert Systems with Applications, vol. 182, p. 115284, 2021

    2021

  6. [14]

    Profit or deceit? mitigating pump and dump in defi via graph and contrastive learning,

    C. Wu, J. Chen, J. Li, J. Xu, J. Jia, Y . Hu, Y . Feng, Y . Liu, and Y . Xiang, “Profit or deceit? mitigating pump and dump in defi via graph and contrastive learning,”IEEE Transactions on Information Forensics and Security, 2025

    2025

  7. [15]

    NFTDisk: Visual detection of wash trading in NFT markets,

    X. Wen, Y . Wang, X. Yue, F. Zhu, and M. Zhu, “NFTDisk: Visual detection of wash trading in NFT markets,” inProceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 2023, pp. 1–15

    2023

  8. [16]

    Bubblemaps: The onchain intelligence layer,

    Bubblemaps, “Bubblemaps: The onchain intelligence layer,” https:// bubblemaps.io/, 2026, accessed: 2026-04-19

    2026

  9. [17]

    Knowledge generation model for visual analytics,

    D. Sacha, A. Stoffel, F. Stoffel, B. C. Kwon, G. Ellis, and D. A. Keim, “Knowledge generation model for visual analytics,”IEEE Transactions on Visualization and Computer Graphics, vol. 20, no. 12, pp. 1604– 1613, 2014

    2014

  10. [18]

    Application of benford’s law on cryptocur- rencies,

    J. Vi ˇciˇc and A. To ˇsi´c, “Application of benford’s law on cryptocur- rencies,”Journal of Theoretical and Applied Electronic Commerce Research, vol. 17, no. 1, pp. 313–326, 2022

    2022

  11. [19]

    Detecting pump-and-dumps with crypto- assets: Dealing with imbalanced datasets and insiders’ anticipated pur- chases,

    D. Fantazzini and Y . Xiao, “Detecting pump-and-dumps with crypto- assets: Dealing with imbalanced datasets and insiders’ anticipated pur- chases,”Econometrics, vol. 11, no. 3, p. 22, 2023

    2023

  12. [20]

    Machine learning-based detection of pump- and-dump schemes in real-time,

    M. Bolz, K. Brundler, L. Kane, P. Patsias, L. Tessendorf, K. Gogol, T. Kim, and C. Tessone, “Machine learning-based detection of pump- and-dump schemes in real-time,”arXiv preprint arXiv:2412.18848, 2024

    2024

  13. [21]

    Visualization of blockchain data: a systematic review,

    N. Tovanich, N. Heulot, J.-D. Fekete, and P. Isenberg, “Visualization of blockchain data: a systematic review,”IEEE Transactions on Visualiza- tion and Computer Graphics, vol. 27, no. 7, pp. 3135–3152, 2019

    2019

  14. [22]

    Blockchain data analytics: Review and challenges,

    R. Mafrur, “Blockchain data analytics: Review and challenges,”arXiv preprint arXiv:2503.09165, 2025

    2025 arXiv

  15. [23]

    Go with the-bitcoin-flow, with visual analytics,

    S. Bistarelli and F. Santini, “Go with the-bitcoin-flow, with visual analytics,” inProceedings of the 12th International Conference on Availability, Reliability and Security, 2017, pp. 1–6

    2017

  16. [24]

    Envisage: Towards expressive visual graph querying,

    X. Wen, Q. Fu, S. Han, Y . Guo, J. K. Liu, and Y . Wan, “Envisage: Towards expressive visual graph querying,”IEEE Transactions on Visu- alization and Computer Graphics, 2025

    2025

  17. [25]

    Code will tell: Visual identification of ponzi schemes on ethereum,

    X. Wen, K. S. Yeo, Y . Wang, L. Cheng, F. Zhu, and M. Zhu, “Code will tell: Visual identification of ponzi schemes on ethereum,” inExtended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems, 2023, pp. 1–6

    2023

  18. [26]

    Ponzilens+: Visualizing bytecode actions for smart ponzi scheme iden- tification,

    X. Wen, T. D. Nguyen, S. Ruan, Q. Shen, J. Sun, F. Zhu, and Y . Wang, “Ponzilens+: Visualizing bytecode actions for smart ponzi scheme iden- tification,”IEEE Transactions on Visualization and Computer Graphics, 2024

    2024

  19. [27]

    Tendrils of crime: Visualiz- ing the diffusion of stolen bitcoins,

    M. Ahmed, I. Shumailov, and R. Anderson, “Tendrils of crime: Visualiz- ing the diffusion of stolen bitcoins,” inProceedings of the International Workshop on Graphical Models for Security. Springer, 2018, pp. 1–12

    2018

  20. [28]

    Supoolvisor: a visual analytics system for mining pool surveillance,

    J.-z. Xia, Y .-h. Zhang, H. Ye, Y . Wang, G. Jiang, Y . Zhao, C. Xie, X.-y. Kui, S.-h. Liao, and W.-p. Wang, “Supoolvisor: a visual analytics system for mining pool surveillance,”Frontiers of Information Technology & Electronic Engineering, vol. 21, no. 4, pp. 507–523, 2020

    2020

  21. [29]

    Bitconeview: visualization of flows in the bitcoin transaction graph,

    G. Di Battista, V . Di Donato, M. Patrignani, M. Pizzonia, V . Roselli, and R. Tamassia, “Bitconeview: visualization of flows in the bitcoin transaction graph,” inProceedings of the 2015 IEEE Symposium on Visualization for Cyber Security (VizSec). IEEE, 2015, pp. 1–8

    2015

  22. [30]

    Bitvis: An interactive visualization system for bitcoin accounts analysis,

    Y . Sun, H. Xiong, S. M. Yiu, and K. Y . Lam, “Bitvis: An interactive visualization system for bitcoin accounts analysis,” inProceedings of the 2019 Crypto Valley Conference on Blockchain Technology (CVCBT). IEEE, 2019, pp. 21–25

    2019

  23. [31]

    Bitanalysis: A visualization system for bitcoin wallet investiga- tion,

    ——, “Bitanalysis: A visualization system for bitcoin wallet investiga- tion,”IEEE Transactions on Big Data, vol. 9, no. 2, pp. 621–636, 2022

    2022

  24. [32]

    BitConduite: Visualizing and analyzing activity on the Bitcoin network,

    C. Kinkeldey, J.-D. Fekete, and P. Isenberg, “BitConduite: Visualizing and analyzing activity on the Bitcoin network,” inProceedings of EuroVis 2017 - Eurographics Conference on Visualization, Posters Track, 2017, p. 3

    2017

  25. [33]

    Bitextract: Interactive visualization for extracting bitcoin exchange in- telligence,

    X. Yue, X. Shu, X. Zhu, X. Du, Z. Yu, D. Papadopoulos, and S. Liu, “Bitextract: Interactive visualization for extracting bitcoin exchange in- telligence,”IEEE Transactions on Visualization and Computer Graphics, vol. 25, no. 1, pp. 162–171, 2018

    2018

  26. [34]

    Visualizing dynamic bitcoin transaction patterns,

    D. McGinn, D. Birch, D. Akroyd, M. Molina-Solana, Y . Guo, and W. J. Knottenbelt, “Visualizing dynamic bitcoin transaction patterns,” Big Data, vol. 4, no. 2, pp. 109–119, 2016

    2016

  27. [35]

    A review on large language models for visual analytics,

    N. S. Agarwal and S. K. Sonbhadra, “A review on large language models for visual analytics,”arXiv preprint arXiv:2503.15176, 2025

    2025 arXiv

  28. [36]

    Generative ai for visualization: State of the art and future directions,

    Y . Ye, J. Hao, Y . Hou, Z. Wang, S. Xiao, Y . Luo, and W. Zeng, “Generative ai for visualization: State of the art and future directions,” Visual Informatics, vol. 8, no. 2, pp. 43–66, 2024

    2024

  29. [37]

    LLM- assisted visual analytics: Opportunities and challenges,

    M. Hutchinson, R. Jianu, A. Slingsby, and P. Madhyastha, “LLM- assisted visual analytics: Opportunities and challenges,”arXiv preprint arXiv:2409.02691, 2024

    2024 arXiv

  30. [38]

    Data Formulator: AI-powered concept-driven visualization authoring,

    C. Wang, J. Thompson, and B. Lee, “Data Formulator: AI-powered concept-driven visualization authoring,”IEEE Transactions on Visual- ization and Computer Graphics, vol. 30, no. 1, pp. 1128–1138, 2024

    2024

  31. [39]

    ChartGPT: Leveraging LLMs to generate charts from abstract natural language,

    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, vol. 31, no. 3, pp. 1731–1745, 2025

    2025

  32. [40]

    Towards natural language interfaces for data visualization: A survey,

    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, vol. 29, no. 6, pp. 3121–3144, 2023

    2023

  33. [41]

    Eviza: A natural language interface for visual analysis,

    V . Setlur, S. E. Battersby, M. Tory, R. Gossweiler, and A. X. Chang, “Eviza: A natural language interface for visual analysis,” inProceed- ings of the 29th Annual Symposium on User Interface Software and Technology. ACM, 2016, pp. 365–377

    2016

  34. [42]

    NL4DV: A toolkit for generating analytic specifications for data visualization from natural language queries,

    A. Narechania, A. Srinivasan, and J. Stasko, “NL4DV: A toolkit for generating analytic specifications for data visualization from natural language queries,”IEEE Transactions on Visualization and Computer Graphics, vol. 27, no. 2, pp. 369–379, 2021

    2021

  35. [43]

    Facilitating conver- sational interaction in natural language interfaces for visualization,

    R. Mitra, A. Narechania, A. Endert, and J. Stasko, “Facilitating conver- sational interaction in natural language interfaces for visualization,” in Proceedings of the 2022 IEEE Visualization and Visual Analytics (VIS). IEEE, 2022, pp. 6–10

    2022

  36. [44]

    Visualizing tree-of-analysis: Facilitating conversational visual analytics for novices,

    F. Qu, T. Tang, Z. Fu, Y . Chen, H. Jia, J. Gao, S. Nurdawulieti, and Y . Wu, “Visualizing tree-of-analysis: Facilitating conversational visual analytics for novices,” inProceedings of the 2026 CHI Conference on Human Factors in Computing Systems. ACM, 2026, pp. 412:1–412:20

    2026

  37. [45]

    LightV A: Lightweight visual analytics with LLM agent-based task planning and execution,

    Y . Zhao, J. Wang, L. Xiang, X. Zhang, Z. Guo, C. Turkay, Y . Zhang, and S. Chen, “LightV A: Lightweight visual analytics with LLM agent-based task planning and execution,”IEEE Transactions on Visualization and Computer Graphics, vol. 31, no. 9, pp. 6162–6177, 2025

    2025

  38. [46]

    InsightPilot: An LLM- empowered automated data exploration system,

    P. Ma, R. Ding, S. Wang, S. Han, and D. Zhang, “InsightPilot: An LLM- empowered automated data exploration system,” inProceedings of the 2023 Conference on Empirical Methods in Natural Language Process- ing: System Demonstrations. Singapore: Association for Computational Lingu...

    2023

  39. [47]

    LEV A: Using large language models to enhance visual analytics,

    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, vol. 31, no. 3, pp. 1830–1847, 2025

    2025

  40. [48]

    ProactiveV A: Proactive visual analytics with LLM-based UI agent,

    Y . Zhao, X. Shu, L. Fan, L. Gao, Y . Zhang, and S. Chen, “ProactiveV A: Proactive visual analytics with LLM-based UI agent,”IEEE Transactions on Visualization and Computer Graphics, vol. 32, no. 1, pp. 451–461, 2026

    2026

  41. [49]

    Bitcoin: A peer-to-peer electronic cash system,

    S. Nakamoto, “Bitcoin: A peer-to-peer electronic cash system,” https: //bitcoin.org/bitcoin.pdf, 2008, accessed: 2026-05-22

    2008

  42. [50]

    A fistful of bitcoins: Characterizing payments among men with no names,

    S. Meiklejohn, M. Pomarole, G. Jordan, K. Levchenko, D. McCoy, G. M. V oelker, and S. Savage, “A fistful of bitcoins: Characterizing payments among men with no names,” inProceedings of the 2013 Internet Measurement Conference, ser. IMC ’13. ACM, 2013, pp. 127–140

    2013

  43. [51]

    Sok: Decentralized exchanges (dex) with automated market maker (amm) protocols,

    J. Xu, K. Paruch, S. Cousaert, and Y . Feng, “Sok: Decentralized exchanges (dex) with automated market maker (amm) protocols,”ACM Computing Surveys, vol. 55, no. 11, pp. 1–50, 2023

    2023

  44. [52]

    Sok: Decentralized finance (defi),

    S. M. Werner, D. Perez, L. Gudgeon, A. Klages-Mundt, D. Harz, and W. J. Knottenbelt, “Sok: Decentralized finance (defi),” inProceedings of the 4th ACM Conference on Advances in Financial Technologies, ser. AFT ’22. ACM, 2022, pp. 30–46. 14

    2022

  45. [54]

    Available: https://arxiv.org/abs/1810.11956

    [Online]. Available: https://arxiv.org/abs/1810.11956

    arXiv

  46. [55]

    Accounts,

    Solana Foundation, “Accounts,” https://solana.com/docs/core/accounts, 2026, accessed: 2026-05-22

    2026

  47. [56]

    Resurrecting address clustering in bitcoin,

    M. M ¨oser, K. Soska, E. Heilman, K. Lee, H. Heffan, S. Srivastava, K. Hogan, J. Hennessey, A. Miller, A. Narayanan, and N. Christin, “Resurrecting address clustering in bitcoin,” inProceedings of Financial Cryptography and Data Security. Springer, 2022, pp. 386–403

    2022

  48. [57]

    Bitcoin address clustering method based on multiple heuristic conditions,

    X. He, B. Yan, G. Yang, J. Su, and X. Chen, “Bitcoin address clustering method based on multiple heuristic conditions,”IET Blockchain, vol. 2, no. 2, pp. 80–94, 2022

    2022

  49. [58]

    Fingerprinting bitcoin entities using money flow representation learning,

    N. Tovanich, M. Soui, N. Heulot, and P. Isenberg, “Fingerprinting bitcoin entities using money flow representation learning,”Applied Network Science, vol. 8, no. 1, p. 50, 2023

    2023

  50. [59]

    Characterizing the ethereum address space,

    J. Payette, S. Schwager, and J. Murphy, “Characterizing the ethereum address space,” Stanford CS229 Project Report, 2017. [Online]. Available: https://cs229.stanford.edu/proj2017/final-reports/5244232.pdf

    2017

  51. [60]

    Address clustering heuristics for ethereum,

    F. Victor, “Address clustering heuristics for ethereum,” inProceedings of Financial Cryptography and Data Security. Springer, 2020, pp. 617–633

    2020

  52. [61]

    Behavior-aware account de-anonymization on ethereum interaction graph,

    J. Zhou, C. Chen, Z. Zheng, J. Wu, Q. Yuan, G. Yu, Z. Yang, and J. Zhao, “Behavior-aware account de-anonymization on ethereum interaction graph,”IEEE Transactions on Information Forensics and Security, vol. 17, pp. 3433–3448, 2022

    2022

  53. [62]

    The economics of cryptocurrency pump and dump schemes,

    N. Gandal, J. T. Hamrick, T. Moore, and T. Oberman, “The economics of cryptocurrency pump and dump schemes,” CEPR Discussion Paper, Tech. Rep. DP13404, 2018. [Online]. Available: https://cepr.org/publications/dp13404

    2018

  54. [63]

    An examination of the cryptocurrency pump-and-dump ecosystem,

    J. T. Hamrick, F. Rouhi, A. Mukherjee, A. Feder, N. Gandal, T. Moore, and M. Vasek, “An examination of the cryptocurrency pump-and-dump ecosystem,”Information Processing & Management, vol. 58, no. 4, p. 102506, 2021

    2021

  55. [64]

    The doge of wall street: Analysis and detection of pump and dump cryptocurrency manipula- tions,

    M. La Morgia, A. Mei, F. Sassi, and J. Stefa, “The doge of wall street: Analysis and detection of pump and dump cryptocurrency manipula- tions,” inProceedings of the ACM Web Conference 2023, ser. WWW ’23. Association for Computing Machinery, 2023, pp. 2168–2177

    2023

  56. [65]

    Defiranger: Detecting price manipulation attacks on defi applications,

    S. Wu, D. Wang, J. He, Y . Zhou, L. Wu, X. Yuan, Q. He, and K. Ren, “Defiranger: Detecting price manipulation attacks on defi applications,” inProceedings of the 32nd USENIX Security Symposium, ser. USENIX Security ’23. USENIX Association, 2023, pp. 2575–2592. [Online]. Availa...

    2023

  57. [66]

    Binance will list act i: The ai prophecy (act) and peanut the squirrel (pnut),

    Binance, “Binance will list act i: The ai prophecy (act) and peanut the squirrel (pnut),” https://www.binance.com/en/support/announcement/ detail/d16d96c136154680a6373225d592bca1, 2024, accessed: 2026-04- 19

    2024

  58. [67]

    Peanut the squirrel (pnut) price chart and market data,

    CoinGecko, “Peanut the squirrel (pnut) price chart and market data,” https://www.coingecko.com/en/coins/peanut-the-squirrel, 2026, accessed: 2026-04-19

    2026

  59. [68]

    Dune analytics,

    Dune Analytics, “Dune analytics,” https://dune.com/, 2026, accessed: 2026-04-19

    2026

  60. [69]

    LIDA: A tool for automatic generation of grammar-agnostic visualizations and infographics using large language models,

    V . Dibia, “LIDA: A tool for automatic generation of grammar-agnostic visualizations and infographics using large language models,” inPro- ceedings of the 61st Annual Meeting of the Association for Computa- tional Linguistics (Volume 3: System Demonstrations), 2023, pp. 113– 126

    2023

  61. [70]

    Smartmlvs: Llm-enabled multiple linked views generation for inter- active visualization,

    T. Qiu, F. Wang, S. Huang, M. Guo, Y . Zhao, J. Li, and S. Chen, “Smartmlvs: Llm-enabled multiple linked views generation for inter- active visualization,” inProceedings of the 2025 IEEE 18th Pacific Visualization Conference (PacificVis). IEEE, 2025, pp. 58–68. Xiaolin Wenis c...

    2025

Pith tools

Reviewed July 14, 2026 · model on record in the stance chip above.