arxiv: 2604.10384 · v1 · submitted 2026-04-11 · 💻 cs.HC

Recognition: unknown

Context-KG: Context-Aware Knowledge Graph Visualization with User Preferences and Ontological Guidance

Rumali Perera , Xiaoqi Wang , Han-Wei Shen

Authors on Pith no claims yet

Pith reviewed 2026-05-10 15:06 UTC · model grok-4.3

classification 💻 cs.HC

keywords knowledge graph visualizationcontext-aware systemsuser preferencesontological guidancelarge language modelssemantic layoutsgraph explorationhuman-computer interaction

0 comments

The pith

Context-KG uses LLMs to extract user preferences from questions and drive ontology-guided layouts for knowledge graphs.

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

Standard KG visualization tools return only direct query matches arranged by force-directed topology and ignore user intent or semantic context. Context-KG instead applies large language models to iteratively pull relevant node types, attributes, and contextual relations from natural language input. These extracted preferences then shape a layout that groups nodes into type-aware regions and supplies placement explanations plus high-level insights. Real-world KG evaluations and a user study report gains in interpretability, relevance, and task success. If correct, this approach would replace generic topological views with query-specific, semantically grounded ones that align better with what users want to learn.

Core claim

Context-KG reframes KG visualization around ontology, context, and user intent by using LLMs to extract preferences and produce tailored, ontology-guided layouts with type-aware regions and high-level insights unavailable in traditional methods.

What carries the argument

LLM-based iterative extraction of user preferences that then control an ontology-guided layout algorithm to form type-aware regions and explain node placement.

If this is right

Node placements gain explicit semantic justifications tied to user intent rather than pure connectivity.
The same KG yields different layouts and insights depending on the natural-language question asked.
Exploration tasks become more efficient because high-level insights are generated automatically beyond direct results.
Visualization systems can adapt dynamically without requiring manual layout adjustments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same preference-extraction step could be reused in other graph visualization tools such as social or biological networks.
Real-time user corrections to the LLM output might create a tighter feedback loop for refining layouts on the fly.
Ontological distance measures may eventually replace or hybridize with force-directed forces in general graph-drawing software.

Load-bearing premise

Large language models can reliably extract accurate user preferences, node types, attributes, and contextual relations from natural language inputs without errors or biases that distort the layout.

What would settle it

A side-by-side user study on the same real-world KGs and questions where participants rate Context-KG visualizations no higher in interpretability, relevance, or task accuracy than standard force-directed layouts.

Figures

Figures reproduced from arXiv: 2604.10384 by Han-Wei Shen, Rumali Perera, Xiaoqi Wang.

**Figure 1.** Figure 1: User query run on the TVCG KG using (a) a traditional force-directed layout and (b) Context-KG, which incorporates dynamically [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗

**Figure 2.** Figure 2: Comparison of visualizations for the query “Given four papers, [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Figure 3: The Context-KG overview: the user question is processed by an LLM to extract preferences, which are combined with KG schema to [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

**Figure 4.** Figure 4: (a) Psuedo-edges for anchoring each node of interest to its cluster [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

**Figure 6.** Figure 6: Iterative context refinement in Context-KG: (a) edge context, (b) [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗

**Figure 7.** Figure 7: User study evaluation results across the nine tasks. (a) Average [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗

**Figure 9.** Figure 9: The DBLP KG queried with "Show the papers published in 1997 [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗

**Figure 8.** Figure 8: Use case showing how the modified query “Find papers published [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗

read the original abstract

Knowledge Graphs (KGs) are increasingly used to represent and explore complex, interconnected data across diverse domains. However, existing KG visualization systems remain limited because they fail to provide the context of user questions. They typically return only the direct query results and arrange them with force-directed layouts by treating the graph as purely topological. Such approaches overlook user preferences, ignore ontological distances and semantics, and provide no explanation for node placement. To address these challenges, we propose Context-KG, a context-aware KG visualization framework. Context-KG reframes KG visualization around ontology, context, and user intent. Using Large Language Models (LLMs), it iteratively extracts user preferences from natural language questions and context descriptions, identifying relevant node types, attributes, and contextual relations. These preferences drive a semantically interpretable, ontology-guided layout that is tailored to each query, producing type-aware regions. Context-KG also generates high-level insights unavailable in traditional methods, opening new avenues for effective KG exploration. Evaluations on real world KGs and a comprehensive user study demonstrate improved interpretability, relevance, and task performance, establishing Context-KG as a new paradigm for KG visualization.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Context-KG uses LLMs to extract user preferences and context from questions then applies ontology rules for type-aware KG layouts and insights, but the claimed gains rest on high-level evaluation statements with no visible details on methods or controls.

read the letter

Context-KG is a framework that uses LLMs to iteratively extract user preferences, relevant node types, attributes, and relations from natural language questions and context descriptions, then drives an ontology-guided layout that creates type-aware regions and produces high-level insights. That combination is the actual new piece. Traditional KG visualization sticks to force-directed layouts on topology alone, so folding in semantic distances and user intent is a reasonable practical move for domains where graphs are large and queries are specific. The paper does a clear job laying out those limitations and showing how the system produces placements that are meant to be interpretable rather than just clustered by connectivity. It also adds insight generation, which standard tools skip. The evaluations on real-world KGs and the user study are said to show gains in interpretability, relevance, and task performance. That direction makes sense for an HCI-focused systems paper. The soft spot is that none of the supporting numbers or controls appear in the description: no baselines named, no participant counts, no task details, no statistical tests, and no discussion of how LLM extraction errors or biases were checked. If the LLM step misidentifies preferences or relations, those mistakes flow straight into the layout and the generated insights, so the central claims depend on an assumption that is not yet shown to hold. This is for visualization and HCI researchers who build or evaluate tools for knowledge-graph exploration in applied domains. A reader working on adaptive interfaces could pick up concrete ideas about preference-driven region layout even if they end up re-implementing parts. I would send it for peer review so the methods and results can be examined in full.

Referee Report

3 major / 3 minor

Summary. The manuscript proposes Context-KG, a context-aware KG visualization framework that uses LLMs to iteratively extract user preferences, relevant node types, attributes, and contextual relations from natural language questions and context descriptions. These extractions inform an ontology-guided, semantically interpretable layout that creates type-aware regions and generates high-level insights unavailable in traditional force-directed approaches. The central claim is that evaluations on real-world KGs and a comprehensive user study demonstrate improvements in interpretability, relevance, and task performance, establishing Context-KG as a new paradigm for KG visualization.

Significance. If the empirical claims hold under scrutiny, the work offers a potentially significant contribution to KG visualization and human-computer interaction by shifting from purely topological layouts to ones incorporating user intent, ontology, and semantics. The LLM-driven extraction mechanism is a novel integration that could enable more personalized and insightful exploration of complex KGs, provided the extraction step proves reliable.

major comments (3)

[Evaluation section] Evaluation section: The abstract and introduction assert that 'evaluations on real world KGs and a comprehensive user study demonstrate improved interpretability, relevance, and task performance,' yet the manuscript supplies no details on the specific real-world KGs tested, baseline visualization systems compared, quantitative metrics (e.g., task completion time, accuracy, or subjective scales), statistical tests, participant count, or exclusion criteria. This omission leaves the central empirical support for the paradigm claim unverifiable and load-bearing.
[Methodology (LLM extraction)] LLM extraction mechanism (core of §3): The framework's layout and insights depend entirely on accurate, unbiased iterative extraction of preferences, node types, attributes, and relations by LLMs from natural language. The paper provides no validation experiments, error-rate analysis, bias checks, or ground-truth comparisons for this step, despite the skeptic note that systematic LLM errors would directly invalidate the ontology-guided regions and generated insights.
[Layout algorithm] Layout algorithm description: The claim of a 'semantically interpretable, ontology-guided layout' with 'type-aware regions' is central, but the manuscript lacks sufficient algorithmic specification (e.g., how ontological distances are quantified and integrated into force-directed or alternative placement rules, or pseudocode) to assess novelty or reproducibility relative to existing semantic layout methods.

minor comments (3)

[Abstract] The abstract could be strengthened by including at least one concrete quantitative result (e.g., percentage improvement or statistical significance) from the user study rather than qualitative assertions.
[Figures] Figure captions and legends should explicitly state what is being compared (e.g., Context-KG vs. baseline) and label all visual elements for clarity without requiring reference to the main text.
[Related Work] Related work section would benefit from explicit comparison tables contrasting Context-KG against at least 3-4 prior KG visualization systems on dimensions such as context awareness and ontology use.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. The comments identify important areas for improving the clarity, rigor, and verifiability of the manuscript. We address each major comment below and commit to revisions that directly respond to the concerns raised.

read point-by-point responses

Referee: [Evaluation section] Evaluation section: The abstract and introduction assert that 'evaluations on real world KGs and a comprehensive user study demonstrate improved interpretability, relevance, and task performance,' yet the manuscript supplies no details on the specific real-world KGs tested, baseline visualization systems compared, quantitative metrics (e.g., task completion time, accuracy, or subjective scales), statistical tests, participant count, or exclusion criteria. This omission leaves the central empirical support for the paradigm claim unverifiable and load-bearing.

Authors: We agree that the evaluation section as currently written lacks the level of detail needed for independent verification. The manuscript reports high-level outcomes but does not fully document the experimental setup. In the revised version we will expand this section to specify the real-world KGs used, the baseline systems, the quantitative metrics (task completion time, accuracy, and subjective scales), the statistical tests applied, the participant count, and the exclusion criteria. We will also add tables summarizing the results and make the study protocol and analysis code available as supplementary material. revision: yes
Referee: [Methodology (LLM extraction)] LLM extraction mechanism (core of §3): The framework's layout and insights depend entirely on accurate, unbiased iterative extraction of preferences, node types, attributes, and relations by LLMs from natural language. The paper provides no validation experiments, error-rate analysis, bias checks, or ground-truth comparisons for this step, despite the skeptic note that systematic LLM errors would directly invalidate the ontology-guided regions and generated insights.

Authors: We acknowledge that the reliability of the LLM extraction step is central and that the current manuscript does not provide dedicated validation. We will add a new subsection in the methodology that reports validation experiments, including accuracy and error rates on a manually annotated test set of natural-language inputs, comparisons against expert ground truth, and checks for systematic biases across different LLMs. This addition will allow readers to assess the robustness of the extraction component. revision: yes
Referee: [Layout algorithm] Layout algorithm description: The claim of a 'semantically interpretable, ontology-guided layout' with 'type-aware regions' is central, but the manuscript lacks sufficient algorithmic specification (e.g., how ontological distances are quantified and integrated into force-directed or alternative placement rules, or pseudocode) to assess novelty or reproducibility relative to existing semantic layout methods.

Authors: We agree that the algorithmic description is currently insufficient for reproducibility and for clearly distinguishing the approach from prior semantic layout techniques. In the revision we will provide a detailed specification of how ontological distances are quantified from the ontology hierarchy, how these distances are incorporated as additional forces within the layout, and the precise rules used to form type-aware regions. Pseudocode for the core placement procedure will be included, along with an explicit comparison to existing semantic visualization methods. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical systems proposal with independent evaluation support

full rationale

The paper is a systems description of a KG visualization framework that uses LLMs to extract preferences and applies ontology-guided layouts, with all central claims resting on described real-world KG evaluations and a user study rather than any equations, fitted parameters, self-referential definitions, or derivation chains. No self-citations are invoked as load-bearing uniqueness theorems, no ansatzes are smuggled, and no predictions reduce to inputs by construction. The framework's assumptions about LLM reliability are empirical risks, not circular reductions.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim depends on the untested reliability of LLM preference extraction and the assumed superiority of the ontology-guided layout; these are treated as domain assumptions rather than derived or externally validated in the abstract.

axioms (1)

domain assumption Large Language Models can iteratively extract user preferences, relevant node types, attributes, and contextual relations from natural language questions and context descriptions with sufficient accuracy to produce useful visualizations.
This assumption is invoked as the core mechanism that allows the system to tailor layouts and generate insights beyond direct query results.

invented entities (1)

Context-KG framework no independent evidence
purpose: To deliver context-aware, ontology-guided knowledge graph visualizations with user-preference-driven layouts and high-level insights
The framework is introduced as the proposed solution; its effectiveness is asserted via internal evaluations rather than independent external evidence.

pith-pipeline@v0.9.0 · 5506 in / 1629 out tokens · 76139 ms · 2026-05-10T15:06:40.636767+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

55 extracted references · 11 canonical work pages

[1]

arXiv preprint arXiv:2402.02136 , year=

A. Bodonhelyi, E. Bozkir, S. Yang, E. Kasneci, and G. Kasneci. User intent recognition and satisfaction with large language models: A user study with chatgpt.arXiv preprint arXiv:2402.02136, 2024. 3

work page arXiv 2024
[2]

Bollacker, C

K. Bollacker, C. Evans, P. Paritosh, T. Sturge, and J. Taylor. Freebase: a collaboratively created graph database for structuring human knowledge. InProceedings of the 2008 ACM SIGMOD international conference on Management of data, pp. 1247–1250, 2008. 2

2008
[3]

A. Both, A. Perevalov, and A. Gashkov. Kingvisher–knowledge graph visualizer. InEuropean Semantic Web Conference, pp. 173–177. Springer,
[4]

Bouza and A

A. Bouza and A. Bernstein. (partial) user preference similarity as classification-based model similarity.Semantic Web, 5(1):47–64, 2014. 2

2014
[5]

P. Chen, Q. Wang, and Y . Tian. Exploring entity-level user preference on the knowledge graph for recommender system. InProceedings of the 2022 5th International Conference on Algorithms, Computing and Artificial Intelligence, pp. 1–7, 2022. 3

2022
[6]

A survey on knowledge-oriented retrieval-augmented gener- ation, arXiv preprint arXiv:2503.10677, 2025

M. Cheng, Y . Luo, J. Ouyang, Q. Liu, H. Liu, L. Li et al. A survey on knowledge-oriented retrieval-augmented generation.arXiv preprint arXiv:2503.10677, 2025. 2

work page arXiv 2025
[7]

Davidson and D

R. Davidson and D. Harel. Drawing graphs nicely using simulated an- nealing.ACM Transactions on Graphics (TOG), 15(4):301–331, 1996. 5

1996
[8]

Denisova, E

E. Denisova, E. Tiribilli, A. Luschi, P. Francia, L. Manetti, L. Bocchi et al. Enabling reliable usability assessment and comparative analysis of medical software: a comprehensive framework for multimodal biomedical imaging platforms.Health and Technology, 14(4):671–682, 2024. 7

2024
[9]

P. Eades. A heuristic for graph drawing.Congressus numerantium, 42(11):149–160, 1984. 3

1984
[10]

Ester, H.-P

M. Ester, H.-P. Kriegel, J. Sander, X. Xu, et al. A density-based algorithm for discovering clusters in large spatial databases with noise. Inkdd, vol. 96, pp. 226–231, 1996. 4

1996
[11]

J. Feng, M. Huang, Y . Yang, and X. Zhu. Gake: Graph aware knowledge embedding. InProceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pp. 641–651,

2016
[12]

T. M. Fruchterman and E. M. Reingold. Graph drawing by force-directed placement.Software: Practice and experience, 21(11):1129–1164, 1991. 1, 3, 5

1991
[13]

G. W. Furnas. Generalized fisheye views.Acm Sigchi Bulletin, 17(4):16– 23, 1986. 2

1986
[14]

Guttmann and P

W. Guttmann and P. Höfner. Relational characterisations of paths.Archive of Formal Proofs https://isa-afp. org/entries/Relational_Paths. html, For- mal proof development, 2020. 3

2020
[15]

Y . Han, L. Xu, S. Chen, D. Zou, and C. Lu. Beyond surface struc- ture: A causal assessment of llms’ comprehension ability.arXiv preprint arXiv:2411.19456, 2024. 3

work page arXiv 2024
[16]

Hirsch, J

C. Hirsch, J. Hosking, and J. Grundy. Interactive visualization tools for exploring the semantic graph of large knowledge spaces. 2009. 2

2009
[17]

Huot and E

S. Huot and E. Lecolinet. Focus+ context visualization techniques for displaying large lists with multiple points of interest on small tactile screens. InIFIP Conference on Human-Computer Interaction, pp. 219–
[18]

Jeong, J

C.-U. Jeong, J. Kim, D. Kim, and K.-A. Sohn. Geokg: geometry-aware knowledge graph embedding for gene ontology and genes.Bioinformatics, 41(4):btaf160, 2025. 3

2025
[19]

Kamada, S

T. Kamada, S. Kawai, et al. An algorithm for drawing general undirected graphs.Information processing letters, 31(1):7–15, 1989. 1, 3

1989
[20]

Kemper.Beginning Neo4j

C. Kemper.Beginning Neo4j. Springer, 2015. 1

2015
[21]

Kzeinberg and É

J. Kzeinberg and É. Tardos. Disjoint paths in densely embedded graphs. InProceedings of IEEE 36th Annual Foundations of Computer Science, pp. 52–61. IEEE, 1995. 3

1995
[22]

P.-M. Law, A. Endert, and J. Stasko. What are data insights to professional visualization users? In2020 IEEE visualization conference (VIS), pp. 181–185. IEEE, 2020. 3

2020
[23]

H. Li, G. Appleby, C. D. Brumar, R. Chang, and A. Suh. Knowledge graphs in practice: Characterizing their users, challenges, and visualiza- tion opportunities.IEEE Transactions on Visualization and Computer Graphics, 30(1):584–594, 2023. 2, 3, 5

2023
[24]

Liang, K

S. Liang, K. Stockinger, T. M. De Farias, M. Anisimova, and M. Gil. Querying knowledge graphs in natural language.Journal of big data, 8(1):3, 2021. 3

2021
[25]

R. Likert. A technique for the measurement of attitudes.Archives of psychology, 1932. 7

1932
[26]

Lissandrini, D

M. Lissandrini, D. Mottin, K. Hose, and T. B. Pedersen. Knowledge graph exploration systems: Are we lost? InAnnual Conference on Innovative Data Systems Research, 2022. 2

2022
[27]

G. Liu, Q. Yao, Y . Zhang, and L. Chen. Knowledge-enhanced recommen- dation with user-centric subgraph network. In2024 IEEE 40th Interna- tional Conference on Data Engineering (ICDE), pp. 1269–1281. IEEE,
[28]

Q. Liu, F. Wang, N. Xu, T. L. Yan, T. Meng, and M. Chen. Monotonic paraphrasing improves generalization of language model prompting. In Findings of the Association for Computational Linguistics: EMNLP 2024, pp. 9861–9877, 2024. 3

2024
[29]

X. Liu, S. Shen, B. Li, P. Ma, R. Jiang, Y . Zhang et al. A survey of text-to-sql in the era of llms: Where are we, and where are we going? IEEE Transactions on Knowledge and Data Engineering, 2025. 3

2025
[30]

S. Lv, C. Wang, J. Xiang, Z. Bao, and M. Li. Boosting knowledge graph with diverse-aware intent inference for recommendations.Neural Networks, p. 107914, 2025. 4

2025
[31]

Aref and Faizan Ur Rehman and Mohamed Abdur Rahman and Saleh M

A. Madkour, W. G. Aref, F. U. Rehman, M. A. Rahman, and S. Basalamah. A survey of shortest-path algorithms.arXiv preprint arXiv:1705.02044,

work page arXiv
[32]

McGrath and J

C. McGrath and J. Blythe. Do you see what i want you to see? the effects of motion and spatial layout on viewers’ perceptions of graph structure. Journal of Social Structure, 5(2):2, 2004. 2

2004
[33]

Mishra and S

G. Mishra and S. K. Mohanty. A fast hybrid clustering technique based on local nearest neighbor using minimum spanning tree.Expert Systems with Applications, 132:28–43, 2019. 6

2019
[34]

Nachmanson, R

L. Nachmanson, R. Prutkin, B. Lee, N. H. Riche, A. E. Holroyd, and X. Chen. Graphmaps: Browsing large graphs as interactive maps. In International Symposium on Graph Drawing, pp. 3–15. Springer, 2015. 4

2015
[35]

Nararatwong, N

R. Nararatwong, N. Kertkeidkachorn, and R. Ichise. Knowledge graph visualization: Challenges, framework, and implementation. In2020 IEEE Third International Conference on Artificial Intelligence and Knowledge Engineering (AIKE), pp. 174–178. IEEE, 2020. 2

2020
[36]

Gpt-4 turbo in the openai api

OpenAI. Gpt-4 turbo in the openai api. Accessed on 20 August 2025. 6

2025
[37]

Palacio-Niño and F

J.-O. Palacio-Niño and F. Berzal. Evaluation metrics for unsupervised learning algorithms.arXiv preprint arXiv:1905.05667, 2019. 6

work page arXiv 1905
[38]

J. Pfeffer. Fundamentals of visualizing communication networks.China Communications, 10(3):82–90, 2013. 2

2013
[39]

Z. Qiu, L. Luo, S. Pan, and A. W.-C. Liew. Unveiling user preferences: A knowledge graph and llm-driven approach for conversational recommen- dation.arXiv preprint arXiv:2411.14459, 2024. 2, 3

work page arXiv 2024
[40]

Raissya, F

H. Raissya, F. Darari, and F. J. Ekaputra. Vizkg: A framework for vi- sualizing sparql query results over knowledge graphs. InProc. 6th Int. Workshop Vis. Interact. Ontologies Linked Data, 2021. 2

2021
[41]

Riva and C

D. Riva and C. Rossetti. Visualization of knowledge graphs with em- beddings: an essay on recent trends and methods.arXiv preprint arXiv:2412.05289, 2024. 3

work page arXiv 2024
[42]

Saket, A

B. Saket, A. Endert, and Ç. Demiralp. Task-based effectiveness of basic visualizations.IEEE transactions on visualization and computer graphics, 25(7):2505–2512, 2018. 5

2018
[43]

Saket, P

B. Saket, P. Simonetto, S. Kobourov, and K. Börner. Node, node-link, and node-link-group diagrams: An evaluation.IEEE Transactions on Visualization and Computer Graphics, 20(12):2231–2240, 2014. 2

2014
[44]

Q. Si, Y . Liu, P. Fu, Z. Lin, J. Li, and W. Wang. Learning class- transductive intent representations for zero-shot intent detection.arXiv preprint arXiv:2012.01721, 2020. 2

work page arXiv 2012
[45]

A. S. Spritzer and C. M. D. S. Freitas. Design and evaluation of mag- netviz—a graph visualization tool.IEEE transactions on visualization and computer graphics, 18(5):822–835, 2011. 3

2011
[46]

Tsaknakis, B

I. Tsaknakis, B. Song, S. Gan, D. Kang, A. Garcia, G. Liu et al. Do llms recognize your latent preferences? a benchmark for latent information discovery in personalized interaction.arXiv preprint arXiv:2510.17132,

work page arXiv
[47]

Y . Tu, R. Qiu, and H.-W. Shen. Kg-pre-view: Democratizing a tvcg knowledge graph through visual explorations. In2024 IEEE 17th Pacific Visualization Conference (PacificVis), pp. 162–171. IEEE, 2024. 2, 3

2024
[48]

Y . Tu, X. Wang, R. Qiu, H.-W. Shen, M. Miller, J. Rao et al. An interactive knowledge and learning environment in smart foodsheds.IEEE Computer Graphics and Applications, 43(3):36–47, 2023. 6

2023
[49]

X. Wang, K. Yen, Y . Hu, and H.-W. Shen. Deepgd: A deep learning framework for graph drawing using gnn.IEEE computer graphics and applications, 41(5):32–44, 2021. 3

2021
[50]

X. Wang, K. Yen, Y . Hu, and H.-W. Shen. Smartgd: A gan-based graph drawing framework for diverse aesthetic goals.IEEE Transactions on Visualization and Computer Graphics, 30(8):5666–5678, 2023. 3

2023
[51]

H. Yin, A. Aryani, S. Petrie, A. Nambissan, A. Astudillo, and S. Cao. A rapid review of clustering algorithms.arXiv preprint arXiv:2401.07389,

work page arXiv
[52]

X. Zhao. V olumetric focus+ context visualization techniques. Master’s thesis, State University of New York at Stony Brook, 2013. 2

2013
[53]

Zhong, M

F. Zhong, M. Xue, J. Zhang, F. Zhang, R. Ban, O. Deussen et al. Force- directed graph layouts revisited: a new force based on the t-distribution. IEEE Transactions on Visualization and Computer Graphics, 2023. 3

2023
[54]

A. Zhu. Knowledge graph visualization for understanding ideas.Int J Cross-Discip Subjects Educ (IJCDSE), 3(1), 2013. 2

2013
[55]

Zimmermann and S

E. Zimmermann and S. Bruckner. Multi-focus probes for context- preserving network exploration and interaction in immersive analytics. arXiv preprint arXiv:2507.01140, 2025. 2

work page arXiv 2025