Pith. sign in

REVIEW 5 major objections 6 minor 1 cited by

Evaluating Robustness of LLMs on Crisis-Related Microblogs across Events, Information Types, and Linguistic Features

T0 review · 5 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that GPT-4o and GPT-4 generalize best across disaster types and information types among six tested LLMs, yet all six models underperform on flood-related tweets and on requests-or-urgent-needs messages, and few-shot…

desk verdict A solid, useful benchmark of six LLMs on crisis microblogs, with a real confound in the flood finding that needs a stratified re-analysis. read the letter →

arxiv 2412.10413 v1 pith:CTQMYP5J submitted 2024-12-08 cs.CL cs.AIcs.SI

Muhammad Imran , Abdul Wahab Ziaullah , Kai Chen , Ferda Ofli This is my paper · ORCID
classification cs.CLcs.AIcs.SI
keywords largelanguagemodelscrisisinformaticsdisasterresponsemicroblogclassificationfew-shotlearningrobustnessHumAIDsocialmediaanalysis
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

This paper asks whether off-the-shelf large language models can automatically sort crisis-related tweets into humanitarian information categories well enough to be useful during real disasters. It benchmarks three proprietary and three open-source LLMs on 77,196 tweets from 19 real-world disasters, testing both zero-shot and few-shot classification. The authors aim to show that proprietary models, especially GPT-4o and GPT-4, generalize reasonably across disasters and information types, but that all models share weak points: flood-related tweets, non-native English, and the requests-or-urgent-needs class. They also argue that adding few-shot examples brings little consistent gain, which matters for real-time disaster response where labeled examples are scarce. The paper's value is a detailed performance map that tells emergency-response systems where to trust LLM output and where to add human review or extra filtering.

What carries the argument

The central object is the HumAID dataset, containing 77,196 human-annotated crisis tweets from 19 disasters across 11 countries, with a test split of 15,160 tweets labeled into ten information categories (one dropped, leaving nine). The argument is carried by a fixed-prompt, temperature-zero classification protocol applied to six LLMs in zero-shot, one-shot, three-shot, five-shot, and ten-shot settings, with macro-F1 as the headline metric. This protocol isolates the models' pretrained generalization ability rather than their ability to absorb task-specific training, and it gives a common yardstick for comparing disaster types, information types, native versus non-native English, and linguistic features. The few-shot condition adds class-specific example tweets to the prompt, which is the main manipulated variable for probing whether in-context examples help.

What would settle it

Compare GPT-4o's zero-shot macro-F1 on native-English flood events (Maryland floods 2018, Midwestern US floods 2019) against non-native-English flood events (Kerala 2018, Sri Lanka 2017) and against non-flood events from the same language groups, using the per-event results in the paper's Figure 9; if native-English floods score close to other native-English events rather than at the low flood level, the 'flood weakness' claim reduces to the language effect.

Watch

Extended reading notes

Core claim

The paper's central discovery is a performance map of six LLMs on crisis microblogs from the HumAID dataset. GPT-4o and GPT-4 achieve macro-F1 scores around 0.75-0.76 in the zero-shot setting on the full test split, outperforming GPT-3.5 (0.66) and clearly beating the open-source models Llama-2 13B (0.56), Mistral 7B (0.63), and Llama-3 8B (0.53). Across disaster types, earthquakes are handled best and floods worst, with the paper reporting that all models score lowest on flood events. Across information types, requests or urgent needs (RUN) is the persistent weak spot, which the authors describe as below 0.60 F1 for GPT models, and the open-source models also struggle with caution-and-advice messages. The paper further finds that few-shot prompting does not reliably improve performance and sometimes hurts it, and that non-native English tweets plus linguistic features such as typos, long character counts, and mid-sentence hashtags degrade accuracy.

Load-bearing premise

The comparison across disaster types treats 'flood' as the explanatory factor, but two of the four flood events come from countries where English is not the native language, and the paper's own analysis shows all models perform worse on non-native English data, so the flood-specific weakness could be a language confound rather than a property of flood content.

Editorial extensions

If this is right

  • Emergency-response deployments should not count on few-shot examples to close performance gaps, since the paper finds minimal or negative returns from adding shots for most models.
  • Flood-related social media content should be flagged as a high-risk input, because all six models score lowest on it and downstream triage systems will need extra filtering or human review for flood events.
  • The persistent confusion between requests-or-urgent-needs and rescue-or-donation messages means that LLM-based classification alone is not ready to prioritize urgent requests without additional disambiguation.
  • The native versus non-native English analysis implies that LLM performance will be systematically lower for crisis tweets from non-native English-speaking regions, which are common in large-scale disasters.
  • Because proprietary models outperform open-source models on every task and setting tested, organizations relying on open-source 7B-13B models should expect a performance ceiling in this application.

Reading between the lines

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

  • The paper's flood finding may be partly a language-effect artifact: two of the four flood events (Kerala and Sri Lanka) come from non-English-speaking countries, and the paper itself shows all models do worse on non-native English; a controlled comparison of floods across native and non-native English would test whether 'flood' is the real cause or a proxy for language difficulty.
  • The minimal few-shot gains hint that in-domain labeled examples are not the bottleneck; retrieval-augmented generation that pulls event-specific context, or targeted instruction tuning, could be more effective for the requests-or-urgent-needs class than simple shots, though the paper does not test these.
  • The mid-sentence hashtag result suggests a cheap preprocessing intervention: re-locating or stripping mid-sentence hashtags before classification could recover lost accuracy, which is an extension the paper does not evaluate.
  • The paper's benchmark could be extended to multimodal crisis data, such as tweets with images, using vision-language models, a direction the authors themselves list as future work.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper evaluates six LLMs (GPT-3.5, GPT-4, GPT-4o, Llama-2 13B, Llama-3 8B, Mistral 7B) on the HumAID crisis microblog dataset across 19 disasters, using zero- and few-shot classification into nine information classes. It reports macro-F1 by disaster type, information type, native/non-native English countries, and linguistic features, and provides event-wise benchmarks. The headline findings are that proprietary models, especially GPT-4/GPT-4o, generalize better; that floods are the hardest disaster type; that urgent-needs tweets are hard for all models; that few-shot examples yield little gain; and that certain linguistic features (e.g., typos, special characters) hurt performance.

Significance. The study is a useful, clearly-scoped empirical benchmark in an underexplored application area. Its strengths are the use of a public dataset, a fixed prompt template with temperature zero, and coverage of both proprietary and open-source models across multiple shot settings. The finding that GPT-4o does not beat a fine-tuned RoBERTa baseline even with 10 shots is an informative result for practitioners. However, the central comparative claims are undermined by the language/flood confound and by the absence of uncertainty quantification, so the paper currently overstates what the data establish.

major comments (5)
  1. [§4.1, §4.3, Figs. 2 and 5] The conclusion that 'most LLMs face challenges in processing flood-related data' (Abstract) is confounded with the native/non-native English split. The flood group comprises Kerala 2018 and Sri Lanka 2017 (non-native) and Midwestern US 2019 and Maryland 2018 (native). Section 4.3 shows all models perform worse on non-native English tweets, and Figure 9 shows the lowest event-level F1s are for Kerala and Sri Lanka for nearly every model. The paper never compares flood vs. non-flood within the same language group, so the flood deficit could be entirely attributable to non-native English. Please provide the stratified analysis (e.g., native flood vs. native other, non-native flood vs. non-native other) or downgrade the claim.
  2. [§1 and §4.2/Table 4] The Introduction states 'with all GPTs achieving F1 below 0.60' for requests or urgent needs, but Table 4 shows GPT-4 and GPT-4o never fall below 0.66 and 0.70, respectively, in any shot setting; only some GPT-3.5 settings are below 0.60. This is a factual contradiction between the text and the appendix results. The body's 'below 0.75' is also not true for GPT-4o in 5- and 10-shot (0.81, 0.85). Please correct the text and make the claim match the data.
  3. [§4, Figs. 2, 5, 9, Table 3] All comparative statements—e.g., 'GPT-4 and GPT-4o offer better generalizability', 'all models consistently show high performance for earthquakes', 'floods consistently remain the lowest'—are based on point estimates without confidence intervals or significance tests. With only 4-7 events per disaster type, the differences could be noise. Provide bootstrap CIs for event-level F1 or a model with random effects for event, and state whether the observed gaps are statistically reliable. This is needed to support the few-shot 'no improvement' claim as well.
  4. [§4.4, Fig. 8] The hashtag-positioning analysis is reported only as radar charts and qualitative descriptions ('hashtags placed in the middle of messages frequently result in higher error rates'). No numerical F1 values or inferential tests are given for the three positions, and the charts appear to show small and inconsistent differences. Please report the underlying F1 values in a table and test whether position has a reliable effect, or soften the claim.
  5. [§4.4, Tables 1-2, §6] The logistic regression treats individual tweets as independent even though they are nested within events and classes, so the standard errors are likely too small; clustering by event (or a mixed-effects model) is needed before the coefficient significances can be trusted. Moreover, the Conclusion's statement that 'message length, typographical errors, and the presence of special characters' are 'significant challenges' is not supported by the reported p-values: typos have p=0.113 (Mistral) and p=0.077 (GPT-4o), and special characters have p=0.091 and p=0.241. Align the Conclusion with the regression results or re-run the analysis with appropriate inference.
minor comments (6)
  1. [§3.1] The acronym ORI is defined for 'other relevant information' but that class is dropped; either remove the definition or state it is excluded before the definition.
  2. [§4.2] The abbreviation RUE is used for 'requests or urgent needs' in the discussion of Figure 3(b) and the confusion matrix text, but the defined acronym is RUN; please use RUN consistently.
  3. [Figure 8] The caption does not define LM2, LM3, and MST; add a legend or define them as Llama-2 13B, Llama-3 8B, and Mistral 7B, respectively.
  4. [Table 3] Use 'zero-shot' instead of '0-shot' for consistency with the rest of the paper.
  5. [§3.2] The prompt template is shown, but the exact additional instructions used for Llama-2 and Mistral 'to control for verbosity' are not provided; include the full prompts in an appendix for reproducibility.
  6. [§4.5] The text says Figure 9 shows results 'across various few-shot settings' but the heatmap includes zero-shot; change to 'across all shot settings'.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the paper's central claims are empirical F1 measurements on a fixed dataset; self-citations to HumAID and the RoBERTa baseline are not load-bearing for the LLM results.

full rationale

The paper's derivation chain is an empirical benchmark: it takes a fixed test split from the HumAID dataset (Section 3.1), runs six LLMs under zero- and few-shot prompts, and reports F1 scores. The headline findings—flood-related data are hardest, RUN underperforms, few-shot gains are minimal, proprietary models beat open-source—are direct measurements, not quantities derived from fitted parameters or from the cited prior work. The self-citations to HumAID [3] and the RoBERTa baseline (Section 4.5, 'the SOTA supervised baseline (i.e., RoBERTa F1=0.78) as we report in [3]') provide the dataset and a comparison point, but they do not define the LLM predictions or the F1 values being analyzed; the RoBERTa score is an external reference value, not an input that forces any LLM result. The logistic regression in Section 4.4 is post-hoc and uses correctness as the dependent variable with linguistic features as predictors; it is not used to construct the benchmark claims. The flood-versus-non-native-English confound noted by the reader is a real validity concern about causal attribution, but it is a statistical/experimental-design issue, not a circularity in the derivation. No equation, fitted parameter, or self-citation chain reduces the paper's conclusions to its inputs.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central evaluation assumes the crowdsourced HumAID labels are ground truth and that the provided test split is unbiased; it also assumes tweets are independent in the logistic regression despite event-level clustering. No free parameters are hidden in the core F1 benchmark, and no new entities are posited.

assumptions (4)
  • domain assumption Crowdsourced HumAID labels are reliable ground truth for all 19 disaster events and nine information classes.
    Section 3.1 uses the HumAID test split without auditing label quality or inter-annotator agreement; any systematic label noise would bias all model F1 scores.
  • domain assumption Tweets in the logistic regression are independent observations.
    Section 4.4 fits a per-tweet binary regression on correct/incorrect classification but does not cluster standard errors by event, even though tweets from the same disaster are correlated.
  • domain assumption The provided HumAID train/test split is representative of the full disaster distribution.
    Section 3.1 uses the provided test split (N=15,160) without re-sampling; if the split is skewed by event or class, the aggregate F1 comparisons inherit that skew.
  • domain assumption The class definitions in the prompt are sufficient for all six LLMs to produce valid category labels.
    Section 3.2 relies on the prompt definitions; if any model interprets the categories differently, the F1 measures reflect prompt-model mismatch rather than pure model capability.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evaluating Robustness of LLMs on Crisis-Related Microblogs across Events, Information Types, and Linguistic Features." pith.science (2026). https://pith.science/paper/CTQMYP5J

@misc{pith2026241210413,
  author       = {Pith},
  title        = {Pith review of: Evaluating Robustness of LLMs on Crisis-Related Microblogs across Events, Information Types, and Linguistic Features},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CTQMYP5J}},
  note         = {Machine review of arXiv:2412.10413}
}
read the original abstract

The widespread use of microblogging platforms like X (formerly Twitter) during disasters provides real-time information to governments and response authorities. However, the data from these platforms is often noisy, requiring automated methods to filter relevant information. Traditionally, supervised machine learning models have been used, but they lack generalizability. In contrast, Large Language Models (LLMs) show better capabilities in understanding and processing natural language out of the box. This paper provides a detailed analysis of the performance of six well-known LLMs in processing disaster-related social media data from a large-set of real-world events. Our findings indicate that while LLMs, particularly GPT-4o and GPT-4, offer better generalizability across different disasters and information types, most LLMs face challenges in processing flood-related data, show minimal improvement despite the provision of examples (i.e., shots), and struggle to identify critical information categories like urgent requests and needs. Additionally, we examine how various linguistic features affect model performance and highlight LLMs' vulnerabilities against certain features like typos. Lastly, we provide benchmarking results for all events across both zero- and few-shot settings and observe that proprietary models outperform open-source ones in all tasks.

Figures

Figures reproduced from arXiv: 2412.10413 by the authors.

Figure 1
Figure 1. Data distributions for (a) events, (b) information types/classes, (c) disaster types, and (d) [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Performance (F1-scores) of LLMs across disaster types and few-shot settings [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Performance (F1 scores) of LLMs across various information types (i.e., classes) [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Confusion matrices for GPT-4o (left) and Mistral 7B (right) models under the zero-shot [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Performance (F1-scores) of LLMs on native-English-speaking vs. non-English￾speaking countries. LM2=Llama-2 13B, LM3=Llama-3 8B, MST=Mistral 7B [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: Multicollinearity analysis of linguistic features [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Impacts of hashtag positioning on LLMs performance (macro F1-scores). LM2=Llama-2 [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: F1-scores for all proprietary and open-source models for 19 events across all k-shot settings [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Overall performance of proprietary and open [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Detecting Actionable Requests and Offers on Social Media During Crises Using LLMs

    cs.IR 2025-04 conditional novelty 6.0 of 10

    Query-specific retrieval of few-shot examples improves LLM classification of actionable disaster requests and offers on social media compared to static prompting.

Reference graph

Works this paper leans on

36 extracted references · 24 canonical work pages · cited by 1 Pith paper

  1. [1]

    Achiam, S

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023

    arXiv 2023

  2. [2]

    F. Alam, F. Ofli, M. Imran, and M. Aupetit. A twitter tale of three hurricanes: Harvey, irma, and maria. Information Systems for Crisis Response and Management (ISCRAM) , 2018

    work page 2018

  3. [3]

    F. Alam, U. Qazi, M. Imran, and F. Ofli. Humaid: Human-annotated disaster incidents data from twitter with deep learning benchmarks. In Proceedings of the International AAAI Conference on Web and social media, volume 15, pages 933–942, 2021

    2021

  4. [4]

    F. Alam, H. Sajjad, M. Imran, and F. Ofli. Crisisbench: Benchmarking crisis-related social media datasets for humanitarian information processing. In Proceedings of the International AAAI conference on web and social media , volume 15, pages 923–932, 2021

    work page 2021

  5. [5]

    Brown, B

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, et al. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901, 2020

    1901

  6. [6]

    Castillo

    C. Castillo. Big crisis data: social media in disasters and time-critical situations . Cambridge University Press, 2016

    work page 2016

  7. [7]

    Castillo, M

    C. Castillo, M. Mendoza, and B. Poblete. Information credibility on twitter. In Proceedings of the 20th international conference on World wide web , pages 675–684, 2011

    work page 2011

  8. [8]

    Dalvi, M

    F. Dalvi, M. Hasanain, S. Boughorbel, B. Mousi, S. Abdaljalil, N. Nazar, A. Abdelali, S. A. Chowdhury, H. Mubarak, A. Ali, et al. Llmebench: A flexible framework for accelerating llms benchmarking. arXiv preprint arXiv:2308.04945, 2023

    arXiv 2023

Show all 36 references
  1. [9]

    Devlin, M.-W

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding, 2019

    2019

  2. [10]

    Dubey, A

    A. Dubey, A. Jauhri, A. Pandey, A. Kadian, A. Al-Dahle, A. Letman, A. Mathur, A. Schelten, A. Yang, A. Fan, et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783, 2024

    2024 arXiv

  3. [11]

    J. Gama, I. Žliobait ˙e, A. Bifet, M. Pechenizkiy, and A. Bouchachia. A survey on concept drift adaptation. ACM computing surveys (CSUR), 46(4):1–37, 2014. 10

    2014

  4. [12]

    J. Han, Z. Zheng, X.-Z. Lu, K.-Y . Chen, and J.-R. Lin. Quakebert: Accurate classification of social media texts for rapid earthquake impact assessment. arXiv preprint arXiv:2405.06684, 2024

    2024 arXiv

  5. [13]

    X. He, D. Lu, D. Margolin, M. Wang, S. E. Idrissi, and Y .-R. Lin. The signals and noise: actionable information in improvised social media channels during a disaster. In Proceedings of the 2017 ACM on web science conference , pages 33–42, 2017

    2017

  6. [14]

    Huang, K

    B. Huang, K. Bradbury, L. M. Collins, and J. M. Malof. Do Deep Learning Models Generalize to Overhead Imagery from Novel Geographic Domains? The xGD Benchmark Problem. In IGARSS 2020 - 2020 IEEE International Geoscience and Remote Sensing Symposium , pages 1476–1479, Waikoloa,...

    2020

  7. [15]

    Ignat, G

    O. Ignat, G. G. Lakshmy, and R. Mihalcea. Cross-cultural inspiration detection and analysis in real and llm-generated social media data. arXiv preprint arXiv:2404.12933, 2024

    2024 arXiv

  8. [16]

    Imran, C

    M. Imran, C. Castillo, F. Diaz, and S. Vieweg. Processing social media messages in mass emergency: A survey. ACM Computing Surveys (CSUR), 47(4):1–38, 2015

    2015

  9. [17]

    Imran, P

    M. Imran, P. Mitra, and C. Castillo. Twitter as a lifeline: Human-annotated twitter corpora for nlp of crisis-related messages. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC 2016) , Paris, France, may 2016. European Language Re...

    2016

  10. [18]

    A. Q. Jiang, A. Sablayrolles, A. Mensch, C. Bamford, D. S. Chaplot, D. de las Casas, F. Bressand, G. Lengyel, G. Lample, L. Saulnier, L. R. Lavaud, M.-A. Lachaux, P. Stock, T. L. Scao, T. Lavril, T. Wang, T. Lacroix, and W. E. Sayed. Mistral 7b, 2023

    2023

  11. [19]

    Kalluri, W

    T. Kalluri, W. Xu, and M. Chandraker. Geonet: Benchmarking unsupervised adaptation across geographies. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 15368–15379, 2023

    2023

  12. [20]

    P. M. Landwehr, W. Wei, M. Kowalchuck, and K. M. Carley. Using tweets to support disaster planning, warning and response. Safety science, 90:33–47, 2016

    2016

  13. [21]

    Y . Liu, M. Ott, N. Goyal, J. Du, M. Joshi, D. Chen, O. Levy, M. Lewis, L. Zettlemoyer, and V . Stoyanov. Roberta: A robustly optimized bert pretraining approach.ArXiv, abs/1907.11692, 2019

    1907 arXiv

  14. [22]

    Y . Ma, S. Chen, S. Ermon, and D. B. Lobell. Transfer learning in environmental remote sensing. Remote Sensing of Environment, 301:113924, Feb. 2024

    2024

  15. [23]

    V . V . Mihunov, N. S. Lam, L. Zou, Z. Wang, and K. Wang. Use of twitter in disaster rescue: lessons learned from hurricane harvey. International Journal of Digital Earth , 13(12):1454– 1466, 2020

    2020

  16. [24]

    Olteanu, C

    A. Olteanu, C. Castillo, F. Diaz, and S. Vieweg. Crisislex: A lexicon for collecting and filtering microblogged communications in crises. In Proceedings of the international AAAI conference on web and social media , pages 376–385, 2014

    2014

  17. [25]

    Gpt-4o system card

    OpenAI. Gpt-4o system card. https://cdn.openai.com/gpt-4o-system-card.pdf ,

  18. [26]

    H. T. Otal and M. A. Canbaz. Llm-assisted crisis management: Building advanced llm platforms for effective emergency response and public collaboration. arXiv preprint arXiv:2402.10908, 2024

    2024 arXiv

  19. [27]

    V . Sanh, L. Debut, J. Chaumond, and T. Wolf. Distilbert, a distilled version of bert: smaller, faster, cheaper and lighter, 2020

    2020

  20. [28]

    Splendiani and A

    S. Splendiani and A. Capriello. Crisis communication, social media and natural disasters–the use of twitter by local governments during the 2016 italian earthquake. Corporate Communications: An International Journal, 27(3):509–526, 2022. 11

    2016

  21. [29]

    Touvron, L

    H. Touvron, L. Martin, K. Stone, P. Albert, A. Almahairi, Y . Babaei, N. Bashlykov, S. Batra, P. Bhargava, S. Bhosale, D. Bikel, L. Blecher, C. C. Ferrer, M. Chen, G. Cucurull, D. Esiobu, J. Fernandes, J. Fu, W. Fu, B. Fuller, C. Gao, V . Goswami, N. Goyal, A. Hartshorn, S. Ho...

    2023

  22. [30]

    Vieweg, A

    S. Vieweg, A. L. Hughes, K. Starbird, and L. Palen. Microblogging during two natural hazards events: what twitter may contribute to situational awareness. In Proceedings of the SIGCHI conference on human factors in computing systems , pages 1079–1088, 2010

    2010

  23. [31]

    R. S. Wilkho, S. Chang, and N. G. Gharaibeh. Ff-bert: A bert-based ensemble for automated classification of web-based text on flash flood events. Advanced Engineering Informatics , 59:102293, 2024

    2024

  24. [32]

    J. Yang, H. Jin, R. Tang, X. Han, Q. Feng, H. Jiang, S. Zhong, B. Yin, and X. Hu. Harnessing the power of llms in practice: A survey on chatgpt and beyond. ACM Transactions on Knowledge Discovery from Data, 18(6):1–32, 2024

    2024

  25. [33]

    Zhang, K

    Y . Zhang, K. Sharma, L. Du, and Y . Liu. Toward mitigating misinformation and social media manipulation in llm era. In Companion Proceedings of the ACM on Web Conference 2024 , pages 1302–1305, 2024

    2024

  26. [34]

    W. X. Zhao, K. Zhou, J. Li, T. Tang, X. Wang, Y . Hou, Y . Min, B. Zhang, J. Zhang, Z. Dong, et al. A survey of large language models. arXiv preprint arXiv:2303.18223, 2023

    2023 arXiv

  27. [35]

    A. W. Ziaullah, F. Ofli, and M. Imran. Monitoring critical infrastructure facilities during disasters using large language models. arXiv preprint arXiv:2404.14432, 2024. A Class-wise results Tables 4 and 5 present the class-wise results across various shots for all the models....

    2024 arXiv

  28. [2024]

    Accessed on 2024-11-24

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.