pith. sign in

arxiv: 2604.08823 · v1 · submitted 2026-04-09 · 💻 cs.HC

Semantic Zooming and Edge Bundling for Multi-Scale Supply Chain Flow Visualization

Songmao Li , Kaixuan Qu , Keer Sun , Bhargav Limbasia , Luciano Nocera (University of Southern California) This is my paper

Pith reviewed 2026-05-10 16:42 UTC · model grok-4.3

classification 💻 cs.HC
keywords semantic zoomingedge bundlingsupply chain visualizationmulti-scale visualizationvisual analyticsgeographic flowshexagonal aggregationinventory sunbursts
0
0 comments X

The pith

A semantic zooming dashboard adapts bundled flows, heatmaps, and sunbursts to visualize supply chain data across scales without clutter.

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

The paper develops a visual analytics dashboard for supply chain networks that switches representations as users change the zoom level. At the widest views it aggregates and bundles origin-destination flows into simpler lines. At medium scales the view becomes hexagonal areas shaded by flow density. At the closest scales it expands into sunburst diagrams that break down inventory hierarchies. Smooth animated transitions connect the levels so the user never loses context. The authors also modify an existing edge-bundling algorithm to keep geographic routes looking plausible on maps.

Core claim

We present a multi-scale visual analytics dashboard that combines Semantic Zooming with Skeleton-Based Edge Bundling (SBEB). The system dynamically adapts its representation based on zoom level: bundled aggregate flows at the macro-scale, hexagonal density heatmaps at the meso-scale, and hierarchical inventory sunbursts at the micro-scale. Built on Vue3 and Deck.gl, it reduces raw orders to 202 warehouse-to-state flows. We contribute a semantic zoom implementation with animated transitions that unifies edge bundling, hexagonal density aggregation, and hierarchical inventory views into a single interface, and an algorithmic adaptation of SBEB for geographic origin-destination flows that adds

What carries the argument

Semantic zooming that switches between Skeleton-Based Edge Bundling for macro-scale flows, hexagonal density heatmaps for meso-scale density, and hierarchical inventory sunbursts for micro-scale detail.

If this is right

  • The interface aggregates thousands of raw orders down to 202 warehouse-to-state flows.
  • Animated transitions maintain context when moving between macro, meso, and micro representations.
  • Directional-sector clustering and adaptive detour constraints keep bundled routes cartographically plausible.
  • A single interface now supports overview, density, and detail views without switching tools.
  • The system is implemented as an interactive web dashboard using standard libraries.

Where Pith is reading between the lines

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

  • The same semantic-zooming structure could be reused for other geographic flow datasets such as traffic or migration.
  • Without a user study the claim that the encodings improve pattern discovery remains untested.
  • Adding live data streams would turn the dashboard into a real-time supply-chain monitor.
  • The directional clustering technique might generalize to non-geographic networks if direction sectors are redefined.

Load-bearing premise

The chosen encodings at each scale actually reduce clutter and make supply-chain patterns visible to users.

What would settle it

A controlled user study in which participants complete pattern-finding tasks on the same supply-chain dataset using the proposed dashboard versus a conventional single-scale flow map, measuring task accuracy and completion time.

Figures

Figures reproduced from arXiv: 2604.08823 by Bhargav Limbasia, Kaixuan Qu, Keer Sun, Luciano Nocera (University of Southern California), Songmao Li.

Figure 1
Figure 1. Figure 1: The three semantic zoom levels demonstrating variable Level of Detail (LOD). (A) Macro-scale ( [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The Supply Chain Flow Map Interface. The dashboard loads and aggregates 51,371 shipment records in the browser. The control panel (top) allows analysts to toggle Skeleton-Based Edge Bundling, adjust visual parameters (opacity, stroke width), and filter by warehouse or date. The map displays bundled OD flows at macro-scale, supporting tasks T1 and T2. 4. Iterative Attraction. Each edge is subdivided into k … view at source ↗
Figure 3
Figure 3. Figure 3: Unbundled baseline. The same 202 warehouse-to-state flows rendered as straight arcs without SBEB. The overlapping edges from four warehouses produce severe visual clutter, making it difficult to distinguish individual corridors or identify routing anomalies. Compare with the bundled result in [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Bundled result. The same 202 flows after applying our adapted SBEB pipeline. Directional-sector clustering separates flows by warehouse and bearing, while adaptive detour constraints preserve geographic plausibility. High-volume corridors (e.g., California to the East Coast) merge into visible trunk lines, reducing visual clutter and exposing the arterial structure of the distribution network [PITH_FULL_I… view at source ↗
Figure 5
Figure 5. Figure 5: Meso-scale density view (z ≥ 6). Hexagonal bins aggregate delivery destinations, with color encoding the dominant product category (green = Home & Garden, red = Apparel, blue = Electronics) and height encoding order count. The category legend (right) supports cross-regional comparison of demand composition (T3). TABLE I: Semantic zoom level mapping. Each zoom range activates a distinct visual representatio… view at source ↗
Figure 6
Figure 6. Figure 6: Micro-scale warehouse sunburst (z ≥ 10, proximity￾triggered). The three concentric rings correspond to the 3-level product taxonomy (category_lvl1, lvl2, lvl3). Shown here is the California warehouse (WH-CA); analysts can compare sunbursts across warehouses to identify inventory– demand misalignment (T4). hexagonal bin triggers a tooltip with order count and top-k categories sorted by frequency. At micro-s… view at source ↗
read the original abstract

Modern supply chain networks involve spatially distributed flows that become difficult to interpret using traditional visualization techniques, producing visual clutter that obscures actionable patterns. We present a multi-scale visual analytics dashboard that combines Semantic Zooming with Skeleton-Based Edge Bundling (SBEB). The system dynamically adapts its representation based on zoom level: bundled aggregate flows at the macro-scale, hexagonal density heatmaps at the meso-scale, and hierarchical inventory sunbursts at the micro-scale. Built on Vue3 and Deck.gl, it reduces raw orders to 202 warehouse-to-state flows. We contribute (1)a semantic zoom implementation with animated transitions that unifies edge bundling, hexagonal density aggregation, and hierarchical inventory views into a single interface; and (2)an algorithmic adaptation of SBEB for geographic origin-destination flows, introducing directional-sector clustering and adaptive detour constraints to preserve cartographic plausibility.

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.

Referee Report

2 major / 3 minor

Summary. The paper presents a multi-scale visual analytics dashboard for supply chain origin-destination flows that integrates semantic zooming with an adapted Skeleton-Based Edge Bundling (SBEB) method. At macro scale it shows bundled aggregate flows, at meso scale hexagonal density heatmaps, and at micro scale hierarchical inventory sunbursts, with animated transitions between levels. Implemented in Vue3 and Deck.gl, the system reduces raw order data to 202 warehouse-to-state flows. The stated contributions are (1) a unified semantic-zoom interface combining the three visual encodings and (2) algorithmic extensions to SBEB that add directional-sector clustering and adaptive detour constraints to maintain cartographic plausibility.

Significance. If the visual encodings and SBEB adaptations prove effective, the work could supply a practical multi-scale tool for interpreting complex geographic flow data in supply-chain analytics, potentially aiding pattern discovery that single-scale visualizations obscure. The targeted modifications to SBEB for preserving directional and geographic realism constitute a concrete, domain-specific extension rather than a generic reuse of existing bundling techniques. The animated transition mechanism that unifies disparate encodings is a clear engineering contribution.

major comments (2)
  1. [Abstract] Abstract and Contributions: The central claim that the dashboard 'reduces visual clutter' and 'reveals actionable patterns' is asserted without any supporting evaluation—neither quantitative clutter metrics (edge overlap, visual density), user studies, nor comparisons against baselines such as straight-line OD flows or unmodified SBEB. This absence directly undermines the asserted utility of the semantic-zoom system and the two listed contributions.
  2. [Algorithmic Adaptation] SBEB Adaptation description: The directional-sector clustering and adaptive detour constraints are presented only at a high level; the manuscript supplies neither pseudocode, parameter settings, before/after renderings, nor quantitative measures (e.g., average detour length or angular deviation) to demonstrate that cartographic plausibility is preserved. Without such evidence the novelty and correctness of the algorithmic adaptation cannot be assessed.
minor comments (3)
  1. [Abstract] Abstract: spacing errors in the contribution list—'(1)a semantic' should read '(1) a semantic' and '(2)an algorithmic' should read '(2) an algorithmic'.
  2. The manuscript would benefit from explicit references to prior multi-scale flow visualization work (e.g., semantic zooming in network viz or geographic bundling techniques) to situate the claimed novelty.
  3. Clarify the exact aggregation pipeline that reduces raw orders to the 202 warehouse-to-state flows; a short data-processing subsection or diagram would improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the opportunity to respond to the referee's report. We have carefully considered the major comments and provide point-by-point responses below. We plan to make revisions to address the concerns raised.

read point-by-point responses

  1. Referee: [Abstract] Abstract and Contributions: The central claim that the dashboard 'reduces visual clutter' and 'reveals actionable patterns' is asserted without any supporting evaluation—neither quantitative clutter metrics (edge overlap, visual density), user studies, nor comparisons against baselines such as straight-line OD flows or unmodified SBEB. This absence directly undermines the asserted utility of the semantic-zoom system and the two listed contributions.

    Authors: We agree that the manuscript would benefit from supporting evidence for the claims of reduced visual clutter and revealed patterns. The current work focuses on the design and implementation of the system rather than a comprehensive evaluation. In the revised version, we will moderate the language in the abstract and contributions to reflect what is demonstrated through the system description and examples. Additionally, we will include a discussion of potential quantitative metrics and outline plans for future user studies. This addresses the concern without overclaiming in the current submission. revision: partial

  2. Referee: [Algorithmic Adaptation] SBEB Adaptation description: The directional-sector clustering and adaptive detour constraints are presented only at a high level; the manuscript supplies neither pseudocode, parameter settings, before/after renderings, nor quantitative measures (e.g., average detour length or angular deviation) to demonstrate that cartographic plausibility is preserved. Without such evidence the novelty and correctness of the algorithmic adaptation cannot be assessed.

    Authors: We acknowledge that the description of the SBEB adaptations is at a high level in the current manuscript. To improve clarity and allow assessment of the novelty, we will expand this section in the revision. Specifically, we will provide pseudocode for the directional-sector clustering and adaptive detour constraint algorithms, specify the parameter values used in our implementation, include before-and-after renderings of the bundling results, and report quantitative measures such as average detour length and angular deviation where applicable. These additions will substantiate the cartographic plausibility preservation. revision: yes

Circularity Check

0 steps flagged

No significant circularity: descriptive system paper with no derivations or predictions

full rationale

The paper is a descriptive report on a visual analytics dashboard and its SBEB adaptation for supply-chain flows. It presents implementation details, contributions, and high-level algorithmic changes (directional-sector clustering, adaptive detour constraints) but contains no equations, fitted parameters, model outputs, or claimed predictions. No derivation chain exists that could reduce to its own inputs by construction. The absence of user studies or clutter metrics is a validity concern, not circularity. The work is self-contained as a system description.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

This is an applied visualization paper with no mathematical model or derivation. The only implicit assumption is the standard visualization premise that visual clutter in flow maps obscures patterns, which is stated as motivation.

axioms (1)
  • domain assumption Traditional visualization techniques produce visual clutter that obscures actionable patterns in spatially distributed supply chain flows
    Explicitly stated in the opening sentence of the abstract as the problem being solved.

pith-pipeline@v0.9.0 · 5461 in / 1412 out tokens · 78877 ms · 2026-05-10T16:42:11.044020+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

  1. [1]

    Visual analysis of supply network risks: Insights from the electronics industry,

    R. C. Basole and M. A. Bellamy, “Visual analysis of supply network risks: Insights from the electronics industry,”Decision Support Systems, vol. 67, pp. 109–120, 2014

    work page 2014

  2. [2]

    Hierarchical edge bundles: Visualization of adjacency re- lations in hierarchical data,

    D. Holten, “Hierarchical edge bundles: Visualization of adjacency re- lations in hierarchical data,”IEEE Transactions on Visualization and Computer Graphics, vol. 12, no. 5, pp. 741–748, 2006

    work page 2006

  3. [3]

    Multi- echelon supply network analysis with interactive visualization,

    S. K. Singh, M. Jenamani, C. Garg, and H. Alirajpurwala, “Multi- echelon supply network analysis with interactive visualization,” in2019 International Conference on Machine Learning, Big Data, Cloud and Parallel Computing (COMITCon). IEEE, 2019, pp. 481–484

    work page 2019

  4. [4]

    Visualizing spatial economic supply chains to enhance sustainability and resilience,

    Y . Han, S. J. Goetz, and C. Schmidt, “Visualizing spatial economic supply chains to enhance sustainability and resilience,”Sustainability, vol. 13, no. 3, p. 1512, 2021

    work page 2021

  5. [5]

    Pad: An alternative approach to the computer interface,

    K. Perlin and D. Fox, “Pad: An alternative approach to the computer interface,” inProceedings of the 20th Annual Conference on Computer Graphics and Interactive Techniques (SIGGRAPH). ACM, 1993, pp. 57–64. SEMANTIC ZOOMING AND EDGE BUNDLING FOR MULTI-SCALE SUPPLY CHAIN FLOW VISUALIZATION 9

    work page 1993

  6. [6]

    Pad++: A zooming graphical interface for exploring alternate interface physics,

    B. B. Bederson and J. D. Hollan, “Pad++: A zooming graphical interface for exploring alternate interface physics,” inProceedings of the 7th Annual ACM Symposium on User Interface Software and Technology (UIST). ACM, 1994, pp. 17–26

    work page 1994

  7. [7]

    Semantic zooming for ontology graph visualizations,

    V . Wiens, S. Lohmann, and S. Auer, “Semantic zooming for ontology graph visualizations,” inProceedings of the 9th Knowledge Capture Conference, 2017, pp. 1–8

    work page 2017

  8. [8]

    The eyes have it: A task by data type taxonomy for information visualizations,

    B. Shneiderman, “The eyes have it: A task by data type taxonomy for information visualizations,” inProceedings of the 1996 IEEE Symposium on Visual Languages. IEEE, 1996, pp. 336–343

    work page 1996

  9. [9]

    Visualization of interfirm relations in a converging mobile ecosystem,

    R. C. Basole, “Visualization of interfirm relations in a converging mobile ecosystem,”Journal of Information Technology, vol. 24, no. 2, pp. 144– 159, 2009

    work page 2009

  10. [10]

    A review of overview+detail, zooming, and focus+context interfaces,

    A. Cockburn, A. Karlson, and B. B. Bederson, “A review of overview+detail, zooming, and focus+context interfaces,”ACM Com- puting Surveys, vol. 41, no. 1, pp. 1–31, 2009

    work page 2009

  11. [11]

    Hierarchical aggregation for informa- tion visualization: Overview, techniques, and design guidelines,

    N. Elmqvist and J.-D. Fekete, “Hierarchical aggregation for informa- tion visualization: Overview, techniques, and design guidelines,”IEEE Transactions on Visualization and Computer Graphics, vol. 16, no. 3, pp. 439–454, 2010

    work page 2010

  12. [12]

    Skeleton-based edge bundling for graph visualization,

    O. Ersoy, C. Hurter, F. V . Paulovich, G. Cantareira, and A. C. Telea, “Skeleton-based edge bundling for graph visualization,”IEEE Trans- actions on Visualization and Computer Graphics, vol. 17, no. 12, pp. 2364–2373, 2011

    work page 2011

  13. [13]

    Graph bundling by kernel density estimation,

    C. Hurter, O. Ersoy, and A. Telea, “Graph bundling by kernel density estimation,”Computer Graphics Forum, vol. 31, no. 3pt1, pp. 865–874, 2012

    work page 2012

  14. [14]

    State of the art in edge and trail bundling techniques,

    A. Lhuillier, C. Hurter, and A. Telea, “State of the art in edge and trail bundling techniques,”Computer Graphics Forum, vol. 36, no. 3, pp. 619–645, 2017

    work page 2017

  15. [15]

    Visualisation of origins, desti- nations and flows with OD maps,

    J. Wood, J. Dykes, and A. Slingsby, “Visualisation of origins, desti- nations and flows with OD maps,”The Cartographic Journal, vol. 47, no. 2, pp. 117–129, 2010

    work page 2010

  16. [16]

    Munzner,Visualization Analysis and Design

    T. Munzner,Visualization Analysis and Design. CRC Press, 2014

    work page 2014