Pith. sign in

REVIEW 4 major objections 5 minor 33 references

Chronotome: Real-Time Topic Modeling for Streaming Embedding Spaces

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 This paper presents Chronotome, an interactive 3D visualization that combines force-based projection, time-on-Z placement, and per-timestep HDBSCAN clustering to let users watch semantic topics split, persist, and shift in real time.

desk verdict Nice integration of known pieces, but the temporal-evolution claims need layout-stability validation and a concrete cluster-tracking algorithm before they hold up. read the letter →

arxiv 2509.01051 v1 pith:QHKHZOQY submitted 2025-09-01 cs.HC cs.CLcs.CVcs.LG

Matte Lim , Catherine Yeh , Martin Wattenberg , Fernanda Viégas , Panagiotis Michalatos This is my paper · ORCID
classification cs.HCcs.CLcs.CVcs.LG
keywords dynamictopicmodelingembeddingvisualizationforce-directedlayoutstreamingclusteringtemporaldataHDBSCANdimensionalityreductioninteractive
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that temporal semantic structure in streaming embedding data can be made directly visible by encoding time as the vertical axis of a live force-directed layout and re-clustering at every batch. If correct, analysts no longer have to choose between coherent static projections and temporal continuity: they can watch a topic form, persist, fork, and drift while the data is being added. The technique is demonstrated on tweets, paintings, and film frames, and the resulting cluster trajectories track recognizable real-world changes, such as an artist's early realistic works giving way to abstract and geometric styles. The contribution is as much interface as algorithm: because the layout is computed in real time, the visual map itself is the analytic tool.

What carries the argument

The load-bearing mechanism is the incremental spring-force graph in the X-Y plane. Every node pair (i, j) has ideal distance d_ideal(i,j)=1-s(i,j), spring constant k(i,j)=s(i,j), and force F_ij = k(i,j)(d_ideal - d_current), following Hooke's law; a dynamic threshold tau filters edges to keep only semantically meaningful attractions. Node masses grow as m_i = m_0 * beta^(b_current - b_initial) with beta = 1.618, so older nodes resist displacement but are not locked in place. Time is encoded by giving all nodes in the same batch the same Z-coordinate. At each timestep, HDBSCAN (a hierarchical density-based clustering method) runs on the 2D X-Y coordinates, and a parent-tracking rule links eac

What would settle it

Run the same dataset through Chronotome twice with batch order shuffled. If the final X-Y positions and the parent-child cluster chains diverge substantially between runs, the map is encoding insertion order rather than embedding similarity. A complementary check is to compare the final X-Y coordinates to a fresh batch-free multidimensional scaling projection of the same embeddings: if old nodes sit where their addition time puts them instead of where their cosine similarities put them, the central claim fails.

Watch

Extended reading notes

Core claim

Chronotome's central claim is that a spring-force embedding with time batched along Z and HDBSCAN clustering run per timestep yields a stable, rotatable 3D visualization in which topic evolution — cluster persistence, splits, and shifts — is visually legible. The layout sets target distances to 1 minus cosine similarity and spring constants equal to cosine similarity, filters edges through a dynamic threshold tau = mu + (log N / log C) * sigma, and increases older node masses by a factor beta = 1.618 with each new batch, keeping the graph from jumping while still allowing new data to move nodes. Clusters are recomputed at each timestep on the X-Y coordinates and linked to a parent cluster fr

Load-bearing premise

The load-bearing premise is that the incremental spring layout with growing node mass keeps a stable, order-independent map of embedding similarity, so the per-timestep clusters and their parent-child links describe real semantic evolution rather than layout artifacts.

Editorial extensions

If this is right

  • Temporal topic exploration becomes an interactive 3D navigation task rather than a precomputed comparison of static snapshots: users can rotate, play back, and add batches live.
  • Cluster splits and semantic drift can be read directly from visual branching and Z-axis displacement, reducing reliance on separate offline topic-modeling pipelines.
  • Because the method only requires time-stamped embeddings, the same tool can monitor social text, image collections, film frames, or any other streaming multimodal data.
  • Storing clustering and parent links per timestep yields a structured history of cluster evolution, not just a single picture, which could feed downstream analyses.
  • Real-time layout means the tool works as a live monitor: as new data arrives, the evolving map is the current analysis output.

Reading between the lines

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

  • Beyond the paper, the specific beta = 1.618 mass schedule looks like an empirical tuning choice; a fixed or logarithmically growing mass would likely preserve the main behavior, and comparing schedules would test whether the golden ratio is load-bearing.
  • I infer that the temporal story is only trustworthy if the incremental layout is order-independent; shuffling batch order and comparing parent-child cluster chains would directly test whether the 'evolution' reflects semantics or insertion order.
  • The same mechanism could be extended to cluster merges by matching clusters across timesteps on both membership and position, turning the tool into a full dynamic cluster editor with convergence as well as divergence.
  • A natural downstream application is change-point detection: a sudden restructuring of clusters at a given Z-slice could flag moments when the underlying semantic distribution actually shifted.
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. The paper introduces Chronotome, a visualization system that combines force-directed layout in the X-Y plane, time projection along the Z-axis, and per-timestep HDBSCAN clustering to support real-time, interactive exploration of evolving semantic themes in temporally ordered embedding datasets. Data is added in timestamped batches; node masses increase geometrically with batch age to stabilize the layout, and clusters are tracked across timesteps via parent-child relationships. Three use cases are presented: Bill Gates' tweets, Picasso's paintings, and frames from The Grand Budapest Hotel. The central claim is that the tool lets users visually track semantic themes, splits, and persistence in a way that static embeddings make difficult.

Significance. If validated, Chronotome would be a useful addition to temporal embedding visualization, offering a continuous 3D view of both semantic proximity and time. The paper is constructive and describes a plausible pipeline using standard components; it also reports scalability limits honestly in Sec. 3.5. The three cross-modal use cases are concrete and illustrate the intended interactions. However, the central claim depends on two unvalidated assumptions: that the incremental force-based layout is semantically faithful and order-independent enough for HDBSCAN clusters to be meaningful, and that the parent-child cluster links are computed by a reliable, specified algorithm. The current evidence is anecdotal, and the reported frame rates undermine the "real-time" aspect at the dataset sizes used. These are load-bearing gaps that require additional analysis and evaluation.

major comments (4)
  1. [Sec. 3.3 and Sec. 5] The load-bearing assumption that the incremental spring layout preserves semantic geometry independently of arrival order is not tested. The mass update in Sec. 3.3, m_i = m0 * beta^(b_current - b_initial) with beta = 1.618, makes early nodes exponentially resistant to movement. Force-directed embeddings are non-convex; the final layout is generally an order-dependent local optimum, not the metric-MDS solution implied in Sec. 3.1. Because HDBSCAN clusters are computed on these 2D coordinates, the cluster splits and persistence described in Sec. 5 could be layout artifacts. Please add a shuffled-batch-order experiment comparing final layout distances and cluster trajectories, and compare against static embeddings (e.g., UMAP or parametric UMAP) to show that the streaming layout reproduces semantic structure.
  2. [Sec. 3.4 / Sec. 4.2] The parent-child cluster tracking algorithm is promised but never specified. Sec. 3.4 says "Implementation details are provided in Sec. 4.2," but Sec. 4.2 contains only HDBSCAN and labeling descriptions. There is no matching criterion (e.g., membership overlap, centroid proximity, or persistence-based hierarchy), no pseudocode, and no discussion of failure cases. Without this, the Across view, color inheritance, and split visualizations are not reproducible. Please provide the exact algorithm and state its assumptions, especially how it behaves when clusters split or when cluster membership changes between timesteps.
  3. [Sec. 4.2 / Sec. 5] The "real-time" claim is not supported at the reported dataset sizes. Sec. 4.2 reports 30 fps at 200 nodes, 8 fps at 360 nodes, and 1 fps at 900 nodes, while all three datasets in Sec. 5 have 669-1000 points. Thus the full-scale interaction rate is roughly 1 Hz, which contradicts the title and abstract's "real time" framing. Please clarify the maximum interactive scale, report the end-to-end latency of clustering and LLM labeling per timestep, and discuss whether these operations are asynchronous with respect to rendering. If "real time" is intended only for small subsets, state that scope explicitly.
  4. [Sec. 5 / Sec. 6] The evaluation is anecdotal. Sec. 5 presents three narratives, and Sec. 6 concludes that Chronotome surfaces insights "difficult to observe in static embeddings," but no user study, baseline comparison, or quantitative metric is provided. To support the central utility claim, add either a controlled user study with objective tasks (e.g., detect a known cluster split, identify a persistent theme) or quantitative comparisons against static embeddings and dynamic topic models, measuring cluster coherence, cluster stability across time, or temporal smoothness. Without this, the use cases are existence proofs and cannot substantiate the claimed advantage over static visualization.
minor comments (5)
  1. [Sec. 3.1] "This idea is equivalent to metric multidimensional scaling" overstates the relation. A thresholded spring model with repulsive forces and a dynamic cutoff does not minimize the standard MDS stress. Suggest rewording to "inspired by" or "closely related to" metric MDS.
  2. [Sec. 3.1] The dynamic threshold tau = mu + (log N / log C) * sigma is confusing because C is set to ln(21). Clarify the logarithm base and how C was selected. Also note that computing mu and sigma over all edge similarities is O(n^2) and may contribute to the scalability bottleneck.
  3. [Abstract / Sec. 3.4] The abstract says "streaming clustering methods," but Sec. 3.4 simply runs HDBSCAN on the cumulative dataset at each timestep. HDBSCAN is not a streaming algorithm. Clarify whether the clustering is incremental or batch-per-timestep, and adjust the terminology accordingly.
  4. [General] No code or data availability statement is provided. For a visualization tool paper, a link to a live demo or repository would greatly improve reproducibility and allow readers to verify the reported behavior.
  5. [Typos and References] Minor presentation issues: the author list has "Catherine Y eh" with a spurious space; several references contain malformed DOIs (e.g., [3], [10], [26]); the color legend in Fig. 3 uses colored squares that are not legible in print. These should be cleaned up.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: Chronotome is a constructive visualization system; its parameters are tunable constants and its use cases are anecdotal.

full rationale

The paper makes no predictive or derivational claim that could reduce to its inputs. The technique is constructive: it takes precomputed embeddings, projects them with a spring-force model, projects time onto the Z-axis, clusters at each timestep with HDBSCAN, and labels clusters with TF-IDF or an LLM. The free parameters (β=1.618, C=ln(21), τ, HDBSCAN settings) are tuning choices, not quantities fitted to a target result, so none of the outputs are forced by construction. The use cases in Sec. 5 are illustrative narratives written by the system designers, which is weak self-evaluation but not circular reasoning. Sec. 3.4 promises parent-cluster tracking with implementation details deferred to Sec. 4.2, but Sec. 4.2 does not actually specify the matching algorithm; this is an omitted specification and a validity gap, not a circular step. Sec. 3.5 explicitly acknowledges scalability limitations and the lack of cluster merges. The only self-citation in the references (Byron & Wattenberg on stacked graphs) appears in related work and is not load-bearing for the paper's central claim. No step in the derivation chain is equivalent by definition to its own input, and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper introduces no physical or theoretical entities. Its 'new' constructs are interface elements (delta cones, parent clusters) and tuning parameters (beta, C), all of which are either visualization metaphors or free parameters. The central claim rests on standard but unvalidated assumptions about embedding quality, HDBSCAN stability, and layout convergence.

free parameters (5)
  • C (threshold constant) = ln(21) ≈ 3.0445
    In tau = mu + (log N / log C) * sigma, C is 'a tunable constant (currently set to ln(21) based on experimentation)' (Sec. 3.1). It controls how strongly low-similarity edges are pruned, affecting the layout.
  • beta (mass multiplier) = 1.618
    In m_i = m0 * beta^(b_current - b_initial), beta is 'chosen empirically to balance node stability and reactivity' (Sec. 3.3). It determines how quickly old nodes become unresponsive to new data.
  • m0 (initial node mass) = not specified
    The base node mass is never given a numeric value, yet it scales all forces in the layout.
  • HDBSCAN hyperparameters = not specified
    min_cluster_size and min_samples are not reported, although cluster structure is the core output of the tool.
  • Repulsive force magnitude below threshold = not specified
    Sec. 3.1 states that for pairs with similarity below tau 'a high repulsive force is applied,' with no formula or value, so the layout cannot be replicated exactly.
assumptions (5)
  • domain assumption Force-directed energy minimization converges to a layout where Euclidean distances approximately match cosine distances (metric MDS).
    Assumed throughout Sec. 3.1; the paper notes equivalence to metric MDS but does not address local minima or convergence in the streaming setting.
  • domain assumption Pretrained embeddings (all-MiniLM-L6-v2 for text, CLIP for images) encode semantic content that clustering can recover.
    Used in Sec. 4.2 without validation on the specific datasets; topics derived from clusters are assumed to reflect true semantics.
  • domain assumption HDBSCAN applied to 2D X-Y coordinates yields stable, semantically meaningful clusters at each timestep.
    Clustering is the basis for all topic labels and temporal comparisons (Sec. 3.4 and 4.2).
  • ad hoc to paper Parent-child cluster tracking is feasible and correct given monotonic data growth and the clustering output.
    Sec. 3.4 says 'we also track each cluster's parent cluster from the previous timestep,' but no algorithm is provided. The cluster-evolution narrative in Sec. 5 depends entirely on this tracking.
  • ad hoc to paper The dynamic threshold tau = mu + (log N / log C) * sigma produces a useful similarity graph for layout.
    Proposed in Sec. 3.1 with C chosen empirically; no justification, ablation, or sensitivity analysis is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Chronotome: Real-Time Topic Modeling for Streaming Embedding Spaces." pith.science (2026). https://pith.science/paper/QHKHZOQY

@misc{pith2026250901051,
  author       = {Pith},
  title        = {Pith review of: Chronotome: Real-Time Topic Modeling for Streaming Embedding Spaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QHKHZOQY}},
  note         = {Machine review of arXiv:2509.01051}
}
read the original abstract

Many real-world datasets -- from an artist's body of work to a person's social media history -- exhibit meaningful semantic changes over time that are difficult to capture with existing dimensionality reduction methods. To address this gap, we introduce a visualization technique that combines force-based projection and streaming clustering methods to build a spatial-temporal map of embeddings. Applying this technique, we create Chronotome, a tool for interactively exploring evolving themes in time-based data -- in real time. We demonstrate the utility of our approach through use cases on text and image data, showing how it offers a new lens for understanding the aesthetics and semantics of temporal datasets.

Figures

Figures reproduced from arXiv: 2509.01051 by the authors.

Figure 1
Figure 1. To visualize how topics evolve in real time, we create a rotatable embedding space where time is encoded along the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. CHRONOTOME contains two main interface components. (A) The rotatable Embedding Plot allows users to explore topic clusters in real time. Currently, the interface is in Playback mode, where users can visualize how nodes moved over time via the timestep slider. There is a filterable legend on the left, and additional visualization controls on the right. (B) Clicking on a cluster label in the legend opens the Data Gall… view at source ↗
Figure 4
Figure 4. Topic clusters derived from The Grand Budapest Hotel film frames visualized in 6-minute intervals using Side View. Switching to Across and Side View, Sam observes many persistent themes throughout Picasso’s works such as ■ “Emotional and ab￾stract artistic representations,” which branches into several child clusters, including “Abstract representations of contemplative fig￾ures” ( [PITH_FULL_IMAGE:figures/full_fig_… view at source ↗
Figures from the paper (1 more)
Figure 3
Figure 3. Figure 3: Topic clusters derived from Bill Gates’ tweets visualized in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 26 canonical work pages

  1. [1]

    M. Ali, A. Alqahtani, M. W. Jones, and X. Xie. Clustering and classifi- cation for time series data in visual analytics: A survey. IEEE Access, 7:181314–181338, 2019. doi: 10.1109/ACCESS.2019.2958551 1

    work page arXiv 2019

  2. [2]

    Artac, M

    M. Artac, M. Jogan, and A. Leonardis. Incremental pca for on-line visual learning and recognition. In 2002 International Conference on Pattern Recognition, vol. 3, pp. 781–784. IEEE, 2002. doi: 10.1109/ TVCG.2008.166 1

    work page 2002

  3. [3]

    Bamler and S

    R. Bamler and S. Mandt. Dynamic word embeddings. In International conference on Machine learning, pp. 380–389. PMLR, 2017. doi: 10. 5555/3305381.3305421 1

    work page arXiv 2017

  4. [4]

    D. M. Blei, A. Y . Ng, and M. I. Jordan. Latent dirichlet allocation. Journal of machine Learning research , 3(Jan):993–1022, 2003. doi: 10.5555/944919.944937 2

    arXiv 2003

  5. [5]

    Boutaleb, J

    A. Boutaleb, J. Picault, and G. Grosjean. Bertrend: Neural topic mod- eling for emerging trends detection. In Proceedings of the Workshop on the Future of Event Detection (FuturED) . Association for Compu- tational Linguistics, 2024. doi: 10.18653/v1/2024.futured-1.1 2

    work page doi:10.18653/v1/2024.futured-1.1 2024

  6. [6]

    Byron and M

    L. Byron and M. Wattenberg. Stacked graphs–geometry & aes- thetics. IEEE transactions on visualization and computer graphics , 14(6):1245–1252, 2008. doi: 10.1109/TVCG.2008.166 1

    work page doi:10.1109/tvcg.2008.166 2008

  7. [7]

    W. Cui, S. Liu, L. Tan, C. Shi, Y . Song, Z. Gao, H. Qu, and X. Tong. Textflow: Towards better understanding of evolving topics in text. IEEE transactions on visualization and computer graphics , 17(12):2412–2421, 2011. doi: 10.1109/TVCG.2011.239 1

    doi:10.1109/tvcg.2011.239 2011

  8. [8]

    Espadoto, R

    M. Espadoto, R. M. Martins, A. Kerren, N. S. T. Hirata, and A. C. Telea. Toward a quantitative survey of dimension reduction tech- niques. IEEE Transactions on Visualization and Computer Graphics , 27(3):2153–2173, 2021. doi: 10.1109/TVCG.2019.2944182 1

    work page arXiv 2021

Show all 33 references
  1. [9]

    Ester, H.-P

    M. Ester, H.-P. Kriegel, J. Sander, X. Xu, et al. A density-based algo- rithm for discovering clusters in large spatial databases with noise. In kdd, vol. 96, pp. 226–231, 1996. doi: 10.5555/3001460.3001507 1

    1996

  2. [10]

    Grootendorst

    M. Grootendorst. Bertopic: Neural topic modeling with a class-based tf-idf procedure. arXiv preprint arXiv:2203.05794 , 2022. doi: 10. 48550/arXiv.2203.05794 2

    2022 doi arXiv

  3. [11]

    Havre, E

    S. Havre, E. Hetzler, P. Whitney, and L. Nowell. Themeriver: Visu- alizing thematic changes in large document collections. IEEE trans- actions on visualization and computer graphics, 8(1):9–20, 2002. doi: 10.1109/2945.981848 1

    2002

  4. [12]

    Hoffman, F

    M. Hoffman, F. Bach, and D. Blei. Online learning for latent dirichlet allocation. Advances in neural information processing systems , 23,

  5. [13]

    Jones, M

    K. Jones, M. A. Runco, C. Dorman, and D. C. Freeland. Influential factors in artist’s lives and themes in their artwork. Creativity Re- search Journal, 10(2-3):221–228, 1997. doi: 10.1080/10400419.1997 .9651221 1

    1997

  6. [14]

    S. G. Kobourov. Spring embedders and force directed graph drawing algorithms. arXiv preprint arXiv:1201.3011 , 2012. doi: 10.48550/ arXiv.1201.3011 2

    2012 doi arXiv

  7. [15]

    J. B. Kruskal. Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika, 29(1):1–27, 1964. doi: 10. 1007/BF02289565 2

    1964

  8. [16]

    S. Liu, M. X. Zhou, S. Pan, Y . Song, W. Qian, W. Cai, and X. Lian. Tiara: Interactive, topic-based visual text summarization and analy- sis. ACM Transactions on Intelligent Systems and Technology (TIST), 3(2):1–28, 2012. doi: 10.1145/2089094.2089101 1, 2

    2012

  9. [17]

    MacQueen

    J. MacQueen. Some methods for classification and analysis of mul- tivariate observations. In Proceedings of the Fifth Berkeley Sympo- sium on Mathematical Statistics and Probability, V olume 1: Statistics, vol. 5, pp. 281–298. University of California press, 1967. 1

    1967

  10. [18]

    Malik, A

    S. Malik, A. Smith, T. Hawes, P. Papadatos, J. Li, C. Dunne, and B. Shneiderman. Topicflow: visualizing topic alignment of twitter data over time. In Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining , ASONAM ’13, p. 7...

    2013

  11. [19]

    McInnes, J

    L. McInnes, J. Healy, S. Astels, et al. Hdbscan: Hierarchical density based clustering. J. Open Source Softw. , 2(11):205, 2017. doi: 10. 21105/joss.00205 1, 3

    2017

  12. [20]

    McInnes, J

    L. McInnes, J. Healy, and J. Melville. Umap: Uniform manifold ap- proximation and projection for dimension reduction. J. Open Source Softw., 3(29):861, 2018. doi: 10.21105/joss.00861 1

    2018 doi

  13. [21]

    Moody, D

    J. Moody, D. McFarland, and S. Bender-deMoll. Dynamic network vi- sualization. American journal of sociology , 110(4):1206–1241, 2005. doi: 10.1086/421509 1

    2005 doi

  14. [22]

    Palamarchuk, L

    D. Palamarchuk, L. Williams, B. Mayer, T. Danielson, R. Faust, L. De- schaine, and C. North. Visualizing temporal topic embeddings with a compass. IEEE Transactions on Visualization and Computer Graph- ics, 2024. doi: 10.1109/TVCG.2024.3456143 1

    2024

  15. [23]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, et al. Learning transferable visual models from natural language supervision. In International conference on machine learning , pp. 8748–8763. PmLR, 2021. doi: 10.4855...

    2021 doi arXiv

  16. [24]

    P. E. Rauber, A. X. Falc ˜ao, and A. C. Telea. Visualizing time- dependent data using dynamic t-sne. InProceedings of the Eurograph- ics / IEEE VGTC Conference on Visualization: Short Papers , EuroVis ’16, p. 73–77. Eurographics Association, Goslar, DEU, 2016. doi: 10 .5555/30...

    2016

  17. [25]

    Rodrigues, C

    N. Rodrigues, C. Schulz, A. Lhuillier, and D. Weiskopf. Cluster-flow parallel coordinates: tracing clusters across subspaces. In Graphics Interface 2020, 2020. doi: 10.20380/GI2020.38 1

    2020 doi

  18. [26]

    Rychlewski

    J. Rychlewski. On hooke’s law. Journal of Applied Mathematics and Mechanics, 48(3):303–314, 1984. doi: 10.1016/0021-8928(84)90137 -0 2

    1984 doi

  19. [27]

    Sainburg, L

    T. Sainburg, L. McInnes, and T. Q. Gentner. Parametric umap embed- dings for representation and semisupervised learning. Neural Compu- tation, 33(11):2881–2907, 2021. doi: 10.1162/neco a 01434 2

    2021 doi

  20. [28]

    Schreck and D

    T. Schreck and D. Keim. Visual analysis of social media data. Com- puter, 46(5):68–75, 2012. doi: 10.1109/MC.2012.430 1

    2012 doi

  21. [29]

    Sparck Jones

    K. Sparck Jones. A statistical interpretation of term specificity and its application in retrieval. Journal of documentation, 28(1):11–21, 1972. doi: 10.5555/106765.106782 2, 3

    1972

  22. [30]

    van den Elzen, D

    S. van den Elzen, D. Holten, J. Blaas, and J. J. van Wijk. Reducing snapshots to points: A visual analytics approach to dynamic network exploration. IEEE transactions on visualization and computer graph- ics, 22(1):1–10, 2015. doi: 10.1109/TVCG.2015.2468078 1, 2

    2015

  23. [31]

    Van der Maaten and G

    L. Van der Maaten and G. Hinton. Visualizing data using t-sne. Jour- nal of machine learning research, 9(11), 2008. 1

    2008

  24. [32]

    Vayansky and S

    I. Vayansky and S. A. Kumar. A review of topic modeling methods.In- formation Systems, 94:101582, 2020. doi: 10.1016/j.is.2020.101582 2

    2020

  25. [2010]

    doi: 10.5555/2997189.2997285 2

Pith tools

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