Unsupervised Multimodal Graph-based Model for Geo-social Analysis

Bernd Resch; Ehsaneddin Jalilian

REVIEW 5 major objections 7 minor 42 references

Unsupervised Multimodal Graph-based Model for Geo-social Analysis

T0 review · 5 major / 7 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read An unsupervised graph model that jointly embeds tweet text and geographic coordinates produces topic clusters that are more semantically coherent and more spatially grounded than existing baselines on four real disaster datasets.

desk verdict A genuinely new multimodal graph architecture with an evaluation that doesn't yet back the headline claim — worth refereeing, but needs real baselines and honest comparisons. read the letter →

arxiv 2512.03063 v2 pith:MPKI62KD submitted 2025-11-26 cs.SI

Ehsaneddin Jalilian , Bernd Resch This is my paper · ORCID

classification cs.SI

keywords geo-socialmediaanalysismultimodalgraphlearningtopicmodelingneuralnetworksunsupervisedclusteringdisastermanagementspatialmulti-headattention

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 authors are trying to establish that semantic and geographic information in geo-tagged social media should be fused inside one end-to-end graph model rather than processed in separate stages. They claim that doing so yields topic clusters that are simultaneously more coherent, more spatially compact, and more interpretable than existing baselines across four disaster datasets. If true, emergency responders and social scientists could extract locally grounded topics from social media without any labeled data. The paper's central mechanism is a composite loss—contrastive, coherence, and alignment—applied to graph embeddings of posts, with two architectural variants: one that mixes modalities in a single graph and one that models them separately and fuses them with attention.

What carries the argument

The load-bearing machinery is a pair of graph constructions over posts-as-nodes: MonoGraph builds one graph whose edges carry either semantic similarity (cosine) or geographic proximity (Haversine distance) and runs a shared graph convolutional encoder; MultiGraph builds separate semantic and geographic kNN graphs, encodes each with its own GCN, then fuses them with multi-head cross-attention in which geographic embeddings act as queries over semantic keys and values. Training is driven by a composite loss: a contrastive term that pulls semantic neighbors together, a coherence term that rewards compact and separated clusters, and an alignment term that keeps nodes near their assigned cluster

What would settle it

Take one of the four datasets, run both proposed models and the sequential baseline at cluster counts 5, 10, 15, and 30, and compute the paper's TQ; also run the external comparison pipeline with its temporal and sentiment features enabled. If the proposed models fail to beat both comparators at any reasonable count—or if the stripped baseline loses only because its auxiliary features were removed—the claim of consistent outperformance is falsified.

Watch

Extended reading notes

Core claim

The paper claims that representing social media posts as nodes in semantic and geographic graphs, then learning embeddings with a graph convolution network under a composite loss, produces topic clusters that beat a sequential embedding-clustering workflow and a previously published multimodal pipeline on all four tested disaster corpora. The MultiGraph variant—which builds separate semantic and geographic k-nearest-neighbor graphs, encodes them with separate GCN branches, and fuses them via multi-head cross-attention in which geographic embeddings query semantic keys and values—scores highest on the paper's topic-quality metric in three of four datasets. The case study of Hurricane Harvey a

Load-bearing premise

The paper's central claim depends on the evaluation being a fair test: the cluster count is fixed by hand (15 for training, 10 for evaluation), the main external baseline is run without its temporal and sentiment modalities, and spatial coherence is never measured on a baseline—if any of those choices is unfavourable, the 'consistently outperform' claim could collapse.

Editorial extensions

If this is right

If the result holds, disaster responders can treat topic clusters as geographically anchored signals—for example, a flood-topic cluster that tracks a storm path—without needing any labeled training data.
MultiGraph outperforms the baselines on three of four events, making modality-separate encoding with attention fusion the stronger default choice when the corpus is linguistically clean; MonoGraph wins on the noisiest, most multilingual corpus.
The ablation shows the contrastive loss is the primary driver: removing it cuts topic quality roughly in half, while coherence and alignment losses act as stabilizers.
Attention fusion beats concatenation and concatenation-plus-MLP fusion, implying that adaptive cross-modal gating—not simply putting features together—is what produces the improvement.
Because the framework is defined at the graph level, the same training objective can be reused when other relation types such as time, sentiment, or image similarity are added as new edge types.

Reading between the lines

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

The geo-as-query attention design implies that geography acts as a learned gate selecting semantic dimensions; a symmetric design with semantic queries would reveal whether the asymmetry itself is essential, which the paper does not test.
The spatial-coherence analysis is run only on the proposed models, so the claim of being more spatially compact than baselines is not yet directly supported; a side-by-side spatial-compactness measurement on the baselines would settle it.
The fixed cluster counts (15 for training, 10 for evaluation) are the most exposed choice: the case study itself shows overlapping weather-related topics, so an adaptive or consensus-based cluster count is a natural extension that could change both interpretability and the reported ranking.
Since the framework is modality-agnostic, the same pipeline could be applied to other paired signals such as image-plus-text or time-plus-location by substituting the two edge types; this is implied by the paper's framing but not tested.

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.

Desk Editor's Note

read the letter

The core idea is genuinely new in the geo-social topic modeling space: jointly embedding tweet text and geolocation through parallel GCN branches, fusing with geo-conditioned attention, and training with a contrastive-plus-coherence-plus-alignment loss. That combination I have not seen before, and it is sensible. The paper also does several things well: it cleanly separates MonoGraph from MultiGraph, the loss ablation is informative (contrastive loss is clearly the main driver), the seed-stability numbers are real, and the Harvey case study shows the intended behavior in qualitative maps.

The soft spots are real and are mostly in the evaluation. The abstract and conclusion claim "consistent outperformance over existing baselines," but the only published baseline, JSTTS from Hanny et al., is run in a stripped configuration for Table 4, and the full-modality results in Table 3 are declared not comparable. Worse, the full-modality JSTTS scores are actually lower than the stripped ones on three of four datasets (Ahr 0.165 vs 0.235, Harvey 0.191 vs 0.222, Napa 0.081 vs 0.143), which directly contradicts the paper's claim that removing sentiment and temporal features explains the pipeline's limited performance. The stress-test note is also correct: Section 5.2.1 explicitly names BERTopic as a baseline, but Table 4 never reports BERTopic — or any standard topic model. The "sequential workfollow" is invented for this paper, so the baseline set is essentially weak and self-defined. Hyperparameters were selected by maximizing TQ on Hurricane Harvey, which is also a reported evaluation dataset, so the Harvey numbers are a selection result, not a prediction — though the ablation and stability sections are honest about variance. Spatial coherence is only ever assessed for the proposed models; Moran's I and Gi* are not computed for any baseline, so the spatial-coherence claim is unsupported. Code and data are not released, which further limits checking.

None of this is fatal to the architecture itself, and it is all addressable. The right fix is a proper comparison set — at minimum LDA, BERTopic, Top2Vec, and ideally the full-modality JSTTS — under the same TQ protocol, plus spatial coherence numbers for all baselines.

Who gets value from this paper: researchers working on multimodal geosocial topic modeling will want to see the architecture and the loss design, but they should not take the benchmark claims at face value. It deserves a serious referee because the method is novel and the experimental gaps can be repaired with reasonable effort. Send it to peer review, but expect a major revision before any accept.

Bottom line: a promising architecture with an overclaimed evaluation. Worth engaging, not worth celebrating yet.

Referee Report

5 major / 7 minor

Summary. The paper proposes an unsupervised graph-based framework that jointly embeds semantic text and geographic coordinates from geotagged social media posts. Two architectures are presented: MonoGraph, which uses a single heterogeneous graph with semantic and geographic edge types, and MultiGraph, which builds separate semantic and geographic graphs and fuses them through multi-head cross-attention. Training uses a composite loss combining contrastive, coherence, and alignment terms, with K-means pseudo-labels providing global guidance. Post hoc spectral clustering (kspect=10) produces topics, which are evaluated on four disaster tweet datasets using intra/inter-cluster similarity and Topic Quality (TQ), defined as NPPMI coherence times topic diversity. A Hurricane Harvey case study adds qualitative topic interpretation and spatial statistics (Moran's I, Gi*, LISA). The authors claim consistent improvement over existing baselines in topic quality, spatial coherence, and interpretability.

Significance. The core idea—jointly optimizing semantic and geographic structure in a single unsupervised GNN—is timely, and the loss decomposition is clearly motivated. The paper provides useful reproducibility details, a domain-agnostic architecture, and a concrete disaster case study. However, as presented, the empirical evidence does not establish the central comparative claim: standard topic-modeling baselines are missing, the one external comparator is used in a configuration whose reported behavior is internally inconsistent, and spatial coherence is never quantified against any baseline. The contribution remains potentially valuable, but the evaluation must be strengthened before the stated claims can be accepted.

major comments (5)

[5.2.1, Table 4] The evaluation protocol names BERTopic as a baseline ('the BERTopic baseline operates solely on semantic representations, reflecting its original formulation'), but no BERTopic result appears in Table 4 or anywhere else. The actual comparators are a self-defined sequential workflow (Section 4.5) and Hanny et al. run without its temporal/sentiment components. The abstract's claim of 'consistently outperform existing baselines' is therefore unsupported against the standard neural and probabilistic topic models cited in the related work. Please add LDA, BERTopic, Top2Vec, and/or ETM under the same TQ protocol, or substantially weaken the claim to the specific comparators used.
[5.3, Tables 3 and 4] There is an internal contradiction in the JSTTS comparison. Table 3 reports full-modality JSTTS TQs (Ahr 0.165, Harvey 0.191, Napa 0.081), while Table 4 reports the stripped 'Hanny et al.' results on the same datasets as 0.235, 0.222, and 0.143—higher in three of four cases. The text explains the stripped pipeline's limited performance as 'likely due to the removal of sentiment and temporal features,' but the data show the opposite direction. Please reconcile the numbers and avoid causal attributions that the table contradicts, or remove the full-modality table if it is not part of a controlled comparison.
[7.1–7.3 and Section 5] The claim of improved 'spatial coherence' is not tested against any baseline. Section 5.1 reports embedding-space intra/inter-cluster similarity, which does not incorporate geographic coordinates. The case study reports Moran's I, Gi*, and LISA maps only for the proposed models; no baseline is run through the same spatial evaluation. To support the stated superiority in spatial coherence, compute a quantitative spatial metric (e.g., Moran's I on topic assignments, spatial silhouette, or distance-based purity) for all baselines under identical clustering and report the comparison.
[6 and Table 4] The sensitivity analysis in Section 6 is performed entirely on Hurricane Harvey, and the 'baseline configuration' (α=0.8, β=0.2, γ=0.1, τ=0.5, kmeans=15) was identified from that analysis. Harvey is then one of the four datasets in the headline TQ results (Table 4). This constitutes tuning on an evaluation dataset. Please either select hyperparameters on a held-out validation set, or report per-dataset sensitivity and demonstrate that the chosen configuration is not tailored to one of the reported test sets.
[3.2, 3.3, 8, and 5.2.2] The fixed cluster counts (kmeans=15 for training, kspect=10 for evaluation) and the fixed top-k keyword count affect the TQ metric. The Discussion (§8) and §7.1 acknowledge that fixed clusters can produce redundant or empty topics. Table 7 varies kmeans only on Harvey and does not vary kspect or the keyword count used in TQ. If the relative ranking of methods changes with kspect, the 'consistent outperformance' claim is fragile. Please include a sensitivity analysis of kspect and keyword k across datasets, or adopt an adaptive/non-parametric cluster count for evaluation.

minor comments (7)

[Abstract, Table 4, §3] Typos: 'MonoGrah' in the abstract, 'Workfollow' in Table 4, 'easyly' in §3, 'coresponding' in §7.1. These should be corrected.
[5.2.2] The phrase 'with k means = 15 in all experiments' confusingly reuses the training-time cluster count for the number of top keywords in TQ. Use a separate variable, e.g., K_words, to avoid ambiguity.
[5.3] The sentence 'the model presented in this paper was originally designed to incorporate sentiment and temporal features' appears to refer to JSTTS/Hanny et al., not the proposed model. As written it misattributes the design and should be rephrased.
[4.4] The claim that strided chunking 'does not affect learned representations or clustering behavior' is asserted but not supported with a quantitative comparison. If this claim is retained, provide the supporting experiment.
[Table 3] The full-modality JSTTS configuration is not described in sufficient detail (clustering procedure, keyword extraction, k, hyperparameters). If this table is retained for contextual reference, these details should be supplied.
[2.1, reference [19]] The text attributes the 'clustering high-dimensional semantic embedding vectors' result to 'Sia et al. [19]', but reference [19] is Wu et al. (2020), a different paper. Please correct the citation.
[4.1 / Data availability] The data availability statement is limited to 'available from the authors upon reasonable request.' For reproducibility, consider a public repository or a more detailed access procedure.

Circularity Check

2 steps flagged · score 6.0 of 10

Partial circularity: the clustering evaluation re-measures the coherence-loss objective itself, and the reported Harvey TQ is selected on Harvey via hyperparameter tuning; the external TQ benchmark is not equation-level circular.

self definitional [Section 3.2 'Loss Design and Training Objective'; Section 5.1 'Clustering Capability Evaluation' (Table 2)]
"At each training epoch, node embeddings are clustered using the K-means algorithm with a fixed number of clusters (k_means = 15)... The resulting cluster assignments are used exclusively to compute the coherence and alignment losses and act as dynamic pseudo-labels... The coherence loss consists of two complementary components: intra-cluster similarity... and inter-cluster similarity... To evaluate the models’ capacity for generating semantically coherent representations... we computed two complementary metrics: intra-cluster similarity... and inter-cluster similarity..."

Table 2's Inter/Intra columns are the same cosine-similarity aggregates that L_coherence directly maximizes/minimizes, with cluster labels taken from K-means applied to the very embeddings being trained. The reported 'input -> MonoGraph/MultiGraph improves clustering' result is therefore the training objective re-measured, not an independent test of clustering quality. The improvement is baked in by construction.
fitted input called prediction [Section 3.2; Section 6 opening and Table 6; Table 4 (Hurricane Harvey row)]
"Unless otherwise stated, we use the baseline configuration α= 0.8, β= 0.2, γ= 0.1, k_means = 15, seed= 42, and τ= 0.5, during training, as identified by the sensitivity analysis in this section. ... All experiments are conducted on the Hurricane Harvey dataset using both the MonoGraph and MultiGraph variants. Performance is evaluated using Topic Quality."

Section 6 states that all sensitivity experiments are run on Hurricane Harvey and evaluated with TQ; the baseline configuration is then chosen because it 'consistently yields the highest TQ.' Those same selected hyperparameters are used to produce the Harvey TQ reported in Table 4 (0.368/0.384). Harvey is thus a selected best case, not an out-of-sample prediction. This does not erase the other three datasets, but it makes part of the 'consistently outperform' claim a tuning artifact on one of the four evaluation events.

full rationale

The central TQ comparison is not equation-level circular: TQ is computed from NPPMI on TF-IDF keywords, a quantity outside the loss, and the improvements over the sequential baseline are real empirical measurements. The Moran's I case study and loss ablations also provide independent evidence. However, the internal clustering evaluation in Section 5.1/Table 2 is a direct re-measurement of the coherence term in L_total: both are cosine-similarity intra/inter-cluster aggregates, with cluster pseudo-labels derived from the embeddings being learned. That portion of the claimed superiority is constructed, not validated. A second, milder loop is the selection of all baseline hyperparameters by TQ on Hurricane Harvey and the subsequent reporting of Harvey as one of the headline results. I do not count the missing BERTopic result or the stripped JSTTS configuration as circularity, since those are evaluation-coverage and correctness concerns rather than derivation loops, but they do strengthen the need to read the 'consistently outperform existing baselines' claim cautiously. Overall: one prediction reduces in part to its training objective, while the external benchmark provides independent content, yielding partial circularity.

Assumptions & free parameters 9 free parameters · 5 assumptions · 0 invented entities

The central empirical claim rests on many hand-set hyperparameters, several tuned on one evaluation dataset, plus an unreported kNN graph parameter; no new physical or conceptual entities are introduced.

free parameters (9)

α (contrastive loss weight) = 0.8
Chosen via sensitivity analysis on Hurricane Harvey (§6.2); part of baseline configuration used for all datasets.
β (coherence loss weight) = 0.2
Chosen via sensitivity analysis on Hurricane Harvey (§6.2); part of baseline configuration.
γ (alignment loss weight) = 0.1
Chosen via sensitivity analysis on Hurricane Harvey (§6.2); part of baseline configuration.
λ (inter-cluster penalty trade-off) = 0.1
Chosen via sensitivity analysis on Hurricane Harvey (§6.2).
τ (contrastive temperature) = 0.5
Chosen via sensitivity analysis on Hurricane Harvey (§6.2, Table 7).
kmeans (training-time cluster count) = 15
Fixed for all datasets; sensitivity on Harvey (§6.2); not justified per event.
kspect (evaluation cluster count) = 10
Fixed for post hoc spectral clustering (§3.3); no sensitivity analysis reported.
k (number of nearest neighbors for semantic/geographic graphs)
Never reported anywhere in the paper; graph topology and all results depend on it.
stride s (chunking parameter)
Stated to be selected by memory constraints; actual values not reported (§4.4).

assumptions (5)

domain assumption SBERT embeddings capture semantic similarity for short, informal tweets.
Relies on Reimers & Gurevych [35]; no validation on the specific corpora used here.
domain assumption GCN message passing along semantic/geographic kNN graphs yields joint embeddings that are clusterable into meaningful topics.
Core inductive bias of the method; not proven beyond the reported experiments.
ad hoc to paper Fixed cluster numbers (15 for training, 10 for evaluation) approximate the true topic structure of all four disaster events.
The paper itself notes that a fixed kmeans may not reflect the true topical complexity of heterogeneous events (Section 8).
domain assumption TQ = TC × TD, with NPPMI and TF-IDF keywords, is a valid measure of topic quality.
Standard-ish metric, but its validity for short noisy disaster tweets is not independently established.
domain assumption Geographic proximity (Haversine distance) is a meaningful relational signal for topic grounding.
Reasonable for geo-social analysis, but the strength of this signal is not quantified against baselines.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Unsupervised Multimodal Graph-based Model for Geo-social Analysis." pith.science (2026). https://pith.science/paper/MPKI62KD

@misc{pith2026251203063,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Multimodal Graph-based Model for Geo-social Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MPKI62KD}},
  note         = {Machine review of arXiv:2512.03063}
}

read the original abstract

The systematic analysis of user-generated social media content, especially when enriched with geospatial context, plays a vital role in domains such as disaster management and public opinion monitoring. Although multimodal approaches have made significant progress, most existing models remain fragmented, processing each modality separately rather than integrating them into a unified end-to-end model. To address this, we propose an unsupervised, multimodal graph-based methodology that jointly embeds semantic and geographic information into a shared representation space. The proposed methodology comprises two architectural paradigms: a mono graph (MonoGrah) model that jointly encodes both modalities, and a multi graph (MultiGraph) model that separately models semantic and geographic relationships and subsequently integrates them through multi-head attention mechanisms. A composite loss, combining contrastive, coherence, and alignment objectives, guides the learning process to produce semantically coherent and spatially compact clusters. Experiments on four real-world disaster datasets demonstrate that our models consistently outperform existing baselines in topic quality, spatial coherence, and interpretability. Inherently domain-independent, the framework can be readily extended to diverse forms of multimodal data and a wide range of downstream analysis tasks.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 3 linked inside Pith

[1]

International Journal of Digital Earth3(3), 231–241 (2010)

Goodchild, M.F., Glennon, J.A.: Crowdsourcing geographic information for dis- aster response: a research frontier. International Journal of Digital Earth3(3), 231–241 (2010)

2010
[2]

during the influenza a h1n1 pandemic

Signorini, A., Segre, A.M., Polgreen, P.M.: The use of twitter to track levels of disease activity and public concern in the u.s. during the influenza a h1n1 pandemic. PLOS ONE6(5), 1–10 (2011)

2011
[3]

Natural Hazards108, 2939–2969 (2021)

Havas, C., Resch, B.: Portability of semantic and spatial–temporal machine learning methods to analyse social media for near-real-time disaster monitoring. Natural Hazards108, 2939–2969 (2021)

2021
[4]

IEEE Transactions on Geoscience and Remote Sensing56(8), 4691–4701 (2018)

Huang, X., Wang, C., Li, Z.: Reconstructing flood inundation probability by enhancing near real-time imagery with real-time gauges and tweets. IEEE Transactions on Geoscience and Remote Sensing56(8), 4691–4701 (2018)

2018
[5]

ISPRS International Journal of Geo-Information10(8), 498 (2021)

Havas, C., Wendlinger, L., Stier, J., Julka, S., Krieger, V., Ferner, C., Petutschnig, A., Granitzer, M., Wegenkittl, S., Resch, B.: Spatio-temporal machine learn- ing analysis of social media data and refugee movement statistics. ISPRS International Journal of Geo-Information10(8), 498 (2021)

2021
[6]

Science Advances9(3), 0199 (2023)

Stolerman, L.M., Clemente, L., Poirier, C., Parag, K.V., Majumder, A., Masyn, S., Resch, B., Santillana, M.: Using digital traces to build prospective and real- time county-level early warning systems to anticipate covid-19 outbreaks in the united states. Science Advances9(3), 0199 (2023)

2023
[7]

Cartography and Geographic Information Science45(4), 362–376 (2018)

Resch, B., Usl¨ ander, F., Havas, C.: Combining machine-learning topic models and spatiotemporal analysis of social media data for disaster footprint and dam- age assessment. Cartography and Geographic Information Science45(4), 362–376 (2018)

2018
[8]

Academy of Management Annals13(2), 586–632 (2019)

Hannigan, T.R., Casasnovas, G.: Topic modeling in management research: Ren- dering new theory from textual data. Academy of Management Annals13(2), 586–632 (2019)

2019

Show all 42 references

[9]

In: Proceedings of the 20th International Conference on World Wide 30 Web, pp

Yin, Z., Cao, L., Han, J., Zhai, C., Huang, T.: A geographical topic model for social media. In: Proceedings of the 20th International Conference on World Wide 30 Web, pp. 247–256 (2011)

2011
[10]

arXiv preprint arXiv:2403.17308 (2024)

Gonz´ alez-Pizarro, C., Carenini, G.: Neural multimodal topic modeling: A com- prehensive evaluation. arXiv preprint arXiv:2403.17308 (2024)

2024 arXiv
[11]

International Journal of Applied Earth Observation and Geoinformation 139, 104540 (2025)

Hanny, D., Resch, B.: Multimodal geoai: An integrated spatio-temporal topic- sentiment model for the analysis of geo-social media posts for disaster manage- ment. International Journal of Applied Earth Observation and Geoinformation 139, 104540 (2025)

2025
[12]

Journal of Computational Social Science 5(1), 97–123 (2022)

Hoyle, A.M., Goel, P., Boyd-Graber, J., Resnik, P.: Content analysis in the age of embeddings: A methodological review. Journal of Computational Social Science 5(1), 97–123 (2022)

2022
[13]

Journal of machine Learning research3(Jan), 993–1022 (2003)

Blei, D.M., Ng, A.Y., Jordan, M.I.: Latent dirichlet allocation. Journal of machine Learning research3(Jan), 993–1022 (2003)

2003
[14]

In: Proceed- ings of the 18th ACM Conference on Information and Knowledge Management

Lin, C., He, Y.: Joint sentiment/topic model for sentiment analysis. In: Proceed- ings of the 18th ACM Conference on Information and Knowledge Management. CIKM ’09, pp. 375–384. Association for Computing Machinery, New York, NY, USA (2009)

2009
[15]

In: Proceedings of the 23rd International Conference on Machine Learning

Blei, D.M., Lafferty, J.D.: Dynamic topic models. In: Proceedings of the 23rd International Conference on Machine Learning. ICML ’06, pp. 113–120. Associa- tion for Computing Machinery, New York, NY, USA (2006)

2006
[16]

In: Proceedings of the 33rd European Conference on Advances in Information Retrieval, pp

Zhao, W.X., Jiang, J., Weng, J., He, J., Lim, E.-P., Yan, H., Li, X.: Comparing twitter and traditional media using topic models. In: Proceedings of the 33rd European Conference on Advances in Information Retrieval, pp. 338–349 (2011)

2011
[17]

In: Proceedings of the First Workshop on Social Media Analytics, pp

Hong, L., Davison, B.D.: Empirical study of topic modeling in twitter. In: Proceedings of the First Workshop on Social Media Analytics, pp. 80–88 (2010)

2010
[18]

Transactions of the Association for Computational Linguistics8, 439–453 (2020)

Dieng, A.B., Ruiz, F.J.R., Blei, D.M.: Topic modeling in embedding spaces. Transactions of the Association for Computational Linguistics8, 439–453 (2020)

2020
[19]

In: Webber, B., Cohn, T., He, Y., Liu, Y

Wu, X., Li, C., Zhu, Y., Miao, Y.: Short text topic modeling with topic distri- bution quantization and negative sampling decoder. In: Webber, B., Cohn, T., He, Y., Liu, Y. (eds.) Proceedings of the 2020 Conference on Empirical Meth- ods in Natural Language Processing (EMNLP),...

2020
[20]

arXiv preprint arXiv:2008.09470

Angelov, D.: Top2Vec: Distributed Representations of Topics. arXiv preprint arXiv:2008.09470. Accessed: 2025-07-03 (2020)

2008 arXiv
[21]

arXiv preprint arXiv:2203.05794

Grootendorst, M.: BERTopic: Neural topic modeling with class-based TF-IDF. arXiv preprint arXiv:2203.05794. Accessed: 2025-07-03 (2022) 31

2025 arXiv
[22]

World Wide Web26(5), 2855–2886 (2023)

Hu, N., Wu, Y., Qi, G., Min, D., Chen, J., Pan, J.Z., Ali, Z.: An empirical study of pre-trained language models in simple knowledge graph question answering. World Wide Web26(5), 2855–2886 (2023)

2023
[23]

In: Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

Yuan, Q., Cong, G., Ma, Z., Sun, A., Thalmann, N.M.: Who, where, when and what: discover spatio-temporal topics for twitter users. In: Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. KDD ’13, pp. 605–613. Association for Comp...

2013
[24]

In: Proceedings of the 21st International Conference on World Wide Web (WWW), pp

Hong, L., Wang, D., Tang, J., Liu, J., Zhang, X.: Discovering geographical topics in the twitter stream. In: Proceedings of the 21st International Conference on World Wide Web (WWW), pp. 769–778 (2012)

2012
[25]

In: International Conference on Web Information Systems Engineering (WISE), pp

Wang, Y., Huang, M., Zhao, L.: Community detection in word co-occurrence networks for topic modeling. In: International Conference on Web Information Systems Engineering (WISE), pp. 399–412 (2015)

2015
[26]

IEEE Transactions on Knowledge and Data Engineering31(11), 2078–2091 (2018)

Zhao, W., Lan, M., Lei, J., Han, J.: Gstm: Graph-based short text topic model- ing. IEEE Transactions on Knowledge and Data Engineering31(11), 2078–2091 (2018)
[27]

Physical Review E69(2), 026113 (2004)

Newman, M.E.J., Girvan, M.: Finding and evaluating community structure in networks. Physical Review E69(2), 026113 (2004)

2004
[28]

In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

Doshi-Velez, F., Wallace, B., Adams, R.: Graph-sparse lda: a topic model with structured sparsity. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 29 (2015)

2015
[29]

In: Webber, B., Cohn, T., He, Y., Liu, Y

Zhou, D., Hu, X., Wang, R.: Neural topic modeling by incorporating document relationship graph. In: Webber, B., Cohn, T., He, Y., Liu, Y. (eds.) Proceedings of the 2020 Conference on Empirical Methods in Natural Language Process- ing (EMNLP), pp. 3790–3796. Association for Com...

2020
[30]

Zhang, L., Liu, J., Yan, Q.: Graph2topic: an opensource topic modeling framework based on sentence embedding and community detection (2023)

2023
[31]

In: Advances in Neural Information Processing Systems (NeurIPS), vol

You, Y., Chen, T., Sui, Y., Chen, T., Wang, Z., Shen, Y.: Graph contrastive learn- ing with augmentations. In: Advances in Neural Information Processing Systems (NeurIPS), vol. 33, pp. 5812–5823 (2020)

2020
[32]

In: Proceedings of the 37th International Conference on Machine Learning (ICML), pp

Hassani, K., Ahmadi, A.H.K.: Contrastive multi-view representation learning on graphs. In: Proceedings of the 37th International Conference on Machine Learning (ICML), pp. 4116–4126 (2020)

2020
[33]

IEEE Transactions on Knowledge and Data Engineering29(1), 17–37 (2016)

Shi, C., Li, Y., Zhang, J., Sun, Y., Yu, P.S.: A survey of heterogeneous information 32 network analysis. IEEE Transactions on Knowledge and Data Engineering29(1), 17–37 (2016)

2016
[34]

IEEE Transactions on Big Data (2018)

Zhang, D., Yin, J., Zhu, X., Zhang, C.: Network representation learning: A survey. IEEE Transactions on Big Data (2018)

2018
[35]

In: Inui, K., Jiang, J., Ng, V., Wan, X

Reimers, N., Gurevych, I.: Sentence-BERT: Sentence embeddings using Siamese BERT-networks. In: Inui, K., Jiang, J., Ng, V., Wan, X. (eds.) Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natur...

2019
[36]

Statistics and Computing17(4), 395–416 (2007)

Luxburg, U.: A tutorial on spectral clustering. Statistics and Computing17(4), 395–416 (2007)

2007
[37]

Seismological Research Letters86(2A), 309–326 (2015)

Brocher, T.M.,et al.: The m6.0 24 august 2014 south napa, california, earthquake. Seismological Research Letters86(2A), 309–326 (2015)

2014
[38]

Accessed: 2025-07-23 (2017)

US Department of Commerce, NOAA: Service Assessment: Hurricane Harvey. Accessed: 2025-07-23 (2017)

2025
[39]

Natural Hazards and Earth System Sciences22(12), 3831–3838 (2022)

Koks, E.E., Ginkel, K.C.H., Marle, M.J.E., Lemnitzer, A.: Brief communication: Critical infrastructure impacts of the 2021 mid-july western european flood event. Natural Hazards and Earth System Sciences22(12), 3831–3838 (2022)

2021
[40]

Scientific Reports14, 1974 (2024)

Cordero, R.R., Feron, S., Damiani, A., Carrasco, J.F., Karas, C., Wang, C., Kraamwinkel, C.T., Beaulieu, A.: Extreme fire weather in chile driven by climate change and el ni˜ no–southern oscillation. Scientific Reports14, 1974 (2024)

1974
[41]

In: Proceedings of GSCL, pp

Bouma, G.: Normalized (pointwise) mutual information in collocation extraction. In: Proceedings of GSCL, pp. 31–40 (2009)

2009
[42]

Prentice Hall PTR, USA (2000) 33

Jurafsky, D., Martin, J.H.: Speech and Language Processing: An Introduc- tion to Natural Language Processing, Computational Linguistics, and Speech Recognition, 1st edn. Prentice Hall PTR, USA (2000) 33

2000

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [1]

International Journal of Digital Earth3(3), 231–241 (2010)

Goodchild, M.F., Glennon, J.A.: Crowdsourcing geographic information for dis- aster response: a research frontier. International Journal of Digital Earth3(3), 231–241 (2010)

2010

[2] [2]

during the influenza a h1n1 pandemic

Signorini, A., Segre, A.M., Polgreen, P.M.: The use of twitter to track levels of disease activity and public concern in the u.s. during the influenza a h1n1 pandemic. PLOS ONE6(5), 1–10 (2011)

2011

[3] [3]

Natural Hazards108, 2939–2969 (2021)

Havas, C., Resch, B.: Portability of semantic and spatial–temporal machine learning methods to analyse social media for near-real-time disaster monitoring. Natural Hazards108, 2939–2969 (2021)

2021

[4] [4]

IEEE Transactions on Geoscience and Remote Sensing56(8), 4691–4701 (2018)

Huang, X., Wang, C., Li, Z.: Reconstructing flood inundation probability by enhancing near real-time imagery with real-time gauges and tweets. IEEE Transactions on Geoscience and Remote Sensing56(8), 4691–4701 (2018)

2018

[5] [5]

ISPRS International Journal of Geo-Information10(8), 498 (2021)

Havas, C., Wendlinger, L., Stier, J., Julka, S., Krieger, V., Ferner, C., Petutschnig, A., Granitzer, M., Wegenkittl, S., Resch, B.: Spatio-temporal machine learn- ing analysis of social media data and refugee movement statistics. ISPRS International Journal of Geo-Information10(8), 498 (2021)

2021

[6] [6]

Science Advances9(3), 0199 (2023)

Stolerman, L.M., Clemente, L., Poirier, C., Parag, K.V., Majumder, A., Masyn, S., Resch, B., Santillana, M.: Using digital traces to build prospective and real- time county-level early warning systems to anticipate covid-19 outbreaks in the united states. Science Advances9(3), 0199 (2023)

2023

[7] [7]

Cartography and Geographic Information Science45(4), 362–376 (2018)

Resch, B., Usl¨ ander, F., Havas, C.: Combining machine-learning topic models and spatiotemporal analysis of social media data for disaster footprint and dam- age assessment. Cartography and Geographic Information Science45(4), 362–376 (2018)

2018

[8] [8]

Academy of Management Annals13(2), 586–632 (2019)

Hannigan, T.R., Casasnovas, G.: Topic modeling in management research: Ren- dering new theory from textual data. Academy of Management Annals13(2), 586–632 (2019)

2019