Pith. sign in

REVIEW 4 major objections 6 minor 29 references

Temporal Analysis of Climate Policy Discourse: Insights from Dynamic Embedded Topic Modeling

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A dynamic topic model of UNFCCC decisions reveals that climate policy discourse has moved from foundational treaties and emissions science to implementation, finance, and capacity building.

desk verdict A clean, reproducible DETM application to UNFCCC decisions with a plausible but under-validated central narrative; the findings are illustrative, not yet demonstrated. read the letter →

arxiv 2507.06435 v1 pith:WHGPUCUX submitted 2025-07-08 cs.CL

Rafiu Adekoya Badekale , Adewale Akinfaderin This is my paper · ORCID
classification cs.CL
keywords dynamicembeddedtopicmodelclimatepolicyUNFCCCdiscourseanalysismodelingtemporalevolutioncomputationalsocialscience
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 applies the Dynamic Embedded Topic Model (DETM) to 28 years of UNFCCC decision documents (1995–2023, omitting 2020) to trace how global climate policy language changes over time. It reports a clear trajectory: early discourse centered on greenhouse gases, international conventions, and institutional groundwork, while recent discourse emphasizes implementation, technical collaboration, capacity building, finance, and global support networks such as the Santiago Network. The authors argue this demonstrates that DETM is a scalable and effective unsupervised tool for analyzing the evolution of policy discourse, in contrast to labor-intensive manual coding. If correct, the finding gives policymakers and researchers a data-driven way to identify when issues enter the agenda, how they evolve, and which themes are gaining momentum.

What carries the argument

The Dynamic Embedded Topic Model (DETM) is the central mechanism: a probabilistic topic model that extends the embedded topic model to time by combining word embeddings with a recurrent neural network, allowing both topic-word distributions and topic prevalence to evolve across time steps. Inference uses a variational autoencoder, producing a beta matrix of topic-word probabilities for each of the 28 yearly time steps. The paper's argument is carried by the temporal word distributions in this matrix, which the authors visualize and align with historical events.

What would settle it

Running the same pipeline with a different number of topics (say 10 or 20) or with different random seeds and checking whether the same shift from 'greenhouse' and 'convention' to 'finance' and 'capacity building' appears, or having domain experts independently label a random sample of the documents and comparing their themes with DETM's topic assignments, would settle whether the observed trajectory is real.

Watch

Extended reading notes

Core claim

The central claim is that DETM, trained on the UNFCCC corpus with five topics, reveals a substantive historical shift in climate policy discourse. In the period 1995–2000, the dominant topic words are 'convention,' 'parties,' 'communications,' 'greenhouse,' and 'gef,' reflecting treaty formation and scientific data collection. From 2001 to 2010 the discourse operationalizes agreements with words like 'project,' 'activities,' 'committee,' and 'review.' Between 2011 and 2016, technical mitigation dominates, with 'storage,' 'geological,' and 'dioxide' marking the carbon capture and storage focus and, by 2015, 'developing,' 'capacity building,' and 'finance' appearing with the Paris Agreement. From 2017 to 2023, 'network,' 'technical,' 'agreement,' 'finance,' and 'santiago' indicate global collaboration and support mechanisms. The paper interprets these trends as evidence of maturation from institutional design to implementation-focused, globally inclusive governance.

Load-bearing premise

The findings rest on the assumption that the five-topic DETM, and the authors' choice to focus on topic 2, faithfully captures the dominant shifts in the UNFCCC discourse; if the hyperparameters or the topic selection are not appropriate, the historical narrative could be an artifact of the model setup rather than a true property of the policy language.

Editorial extensions

If this is right

  • Policymakers can use DETM to identify when specific issues, such as carbon capture and storage or climate finance, enter the discourse and how their prominence changes, informing retrospective evaluation and the timing of new initiatives.
  • Delegates preparing for COP negotiations can align proposals with historical discourse trends and detect emerging topics that are gaining momentum.
  • The approach is scalable to other policy domains, such as public health reports, where the authors suggest it could track shifts in priorities over time.
  • A data-driven measurement of the shift toward implementation, finance, and capacity building supports adaptive, evidence-based climate governance frameworks.

Reading between the lines

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

  • An extension the paper leaves implicit: varying the number of topics (e.g., 10 or 20) and the random seed would test whether the same historical arc from emissions science to finance and capacity building is a stable property of the corpus or an artifact of the five-topic configuration.
  • The method could be applied to national policy documents, such as Nationally Determined Contributions, to test whether the global shift observed in UNFCCC decisions is mirrored at the country level; the paper analyzes only the international corpus.
  • Because DETM captures word distributions but not sentiment or argument structure, combining it with sentiment or argument mining could reveal not just which themes dominate but how they are framed, a question the paper raises but does not answer.
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 / 6 minor

Summary. The paper applies the Dynamic Embedded Topic Model (DETM) to a corpus of UNFCCC decision documents from 1995 to 2023 (excluding 2020), preprocesses the texts into paragraphs, trains a five-topic model, and then selects one topic (topic 2) for close reading based on CCS-related keyword matches. The authors interpret the temporal evolution of topic 2's top words as evidence that global climate policy discourse shifted from early concerns with greenhouse gases and institutional frameworks toward implementation, finance, capacity building, and collaborative mechanisms. The paper positions DETM as a scalable, effective tool for policy discourse analysis and provides code and data on GitHub.

Significance. If the central claim is supported, the paper offers a useful demonstration of a modern topic model for policy corpora and provides a reproducible pipeline. The qualitative alignment with well-known milestones (Kyoto Protocol, Paris Agreement, Santiago Network) is a strength, as is the public release of code and data. However, the paper's contribution is currently an application narrative rather than a validated method: the absence of quantitative evaluation and robustness checks means the main findings could be artifacts of hyperparameter choices. The study is potentially valuable for computational social science but needs substantially more validation to justify the claims in the abstract.

major comments (4)
  1. [Section 3.3 and Figure 2] The paper's central narrative is entirely derived from the temporal evolution of topic 2, which was selected post hoc using CCS-related keywords. Because DETM is a variational model with random initialization and the paper reports only a single run with K=5 and five epochs, the topic-word distributions are not shown to be stable. A different random seed or a different number of topics could produce a different decomposition, altering the timeline in Figure 2. The authors should report topic stability across seeds, vary K (e.g., 3, 5, 7, 10), and show that the qualitative trend in topic 2 persists. Without this, the abstract's claim that DETM 'reveals' the discourse shift is not yet established.
  2. [Section 4, especially 4.1 and 4.3] The paper claims DETM is 'effective' and 'scalable' but reports no quantitative evaluation. There is no topic coherence or perplexity measurement, no comparison with simpler baselines such as LDA, DTM, or a word-frequency trend analysis, and no evaluation of whether the five-topic solution is a good fit to the data. The authors should add standard topic model diagnostics (e.g., normalized pointwise mutual information, held-out perplexity) and at least one baseline comparison to demonstrate that DETM provides added value over static or simpler dynamic topic models.
  3. [Sections 3.2 and 4.4] Section 4.4 acknowledges that DETM is 'sensitive to corpus quality and preprocessing,' but no sensitivity analysis is provided. In particular, the min_df=100 threshold reduces the vocabulary to 960 terms; the authors should test whether the key topic 2 words (e.g., 'santiago', 'geological', 'capacity') survive alternative thresholds (e.g., 50, 150) and preprocessing choices. The paper also does not report validation loss curves or convergence diagnostics beyond a statement that five epochs were 'sufficient,' so the reader cannot assess whether the model has converged.
  4. [Sections 3.3 and 4.1] The paper selects topic 2 out of five and analyzes it in depth, with the other topics relegated to the appendix. Because the claimed discourse shift is based on this single topic, the authors should justify why topic 2 is representative of overall policy discourse and show whether the other topics exhibit similar temporal patterns. If the purpose is to analyze global policy discourse generally, a topic-level prevalence analysis (theta over time) would be more appropriate than focusing on the word distribution of one manually selected topic.
minor comments (6)
  1. [Abstract] The sentence 'A probabilistic model designed to capture the temporal dynamics of topics over time' is a sentence fragment; it should be integrated with the preceding sentence.
  2. [References] References [28] and [29] are listed in the bibliography but do not appear to be cited in the text.
  3. [Figure 2] The text states that six-year intervals were chosen, but the displayed years (e.g., 1995, 2000, 2006, 2010, 2016, 2022) are not consistently six years apart; please clarify the interval rule.
  4. [Section 4.1] The caption for Figure 2 names topic 2 as 'Climate Policy Mechanisms and Mitigation Strategies,' but this label is not derived from the model; it is an interpretive label. Consider indicating which labels are model-derived and which are author-assigned.
  5. [Abstract] The phrase 'we concluded with future directions' is grammatically awkward; consider 'we conclude with future directions'.
  6. [Section 3.1 and Table 1] The corpus statistics in Table 1 describe paragraphs as documents (196,290 train instances). Please clarify in Section 3.1 that the unit of analysis is a paragraph aggregated by year, not a full decision document.

Circularity Check

2 steps flagged · score 4.0 of 10

CCS keyword selection makes the central CCS finding definitional, and the DETM 'effectiveness' claim rests on the same fitted topic-word distributions; external historical alignment keeps the analysis partially grounded.

  1. self definitional [Section 3.3 (Topic Selection); Section 4.1 (2011–2016)]
    "We selected topic 2 based on its relevance to Carbon Capture and Storage (CCS), a key area in climate mitigation. The selection was guided by keyword matches using a predefined set of CCS-related terms (e.g., 'carbon,' 'capture,' 'storage,' 'CCS'). ... Terms like 'storage,' 'geological,' and 'dioxide' signals a focus on Carbon Capture and Storage (CCS) as a mitigation strategy."

    Topic 2 was chosen precisely because its fitted word distribution matched a predefined CCS keyword set containing 'carbon,' 'capture,' and 'storage.' The later finding that topic 2's top words signal 'a focus on Carbon Capture and Storage' is therefore guaranteed by the selection criterion: any topic selected for containing 'storage' will, by construction, display storage-related vocabulary. The 2011–2016 'technical mitigation' phase of the narrative is assembled from the same keywords used to pick the topic, so the CCS observation is a restatement of the input rather than an independent model discovery.

  2. fitted input called prediction [Abstract; Section 4.2]
    "The model reveals shifts from early emphases on greenhouse gases and international conventions to recent focuses on implementation, technical collaboration, capacity building, finance, and global agreements. ... DETM's ability to model latent topic proportions over time enabled us to capture transitions in discourse emphases."

    The 'shifts' are read directly from the fitted beta matrix of the selected topic on the training corpus; no held-out evaluation, baseline comparison (LDA/DTM), topic-stability check, random-seed analysis, or human annotation is used to confirm that the transitions are signal rather than artifacts of K=5, initialization, or five-epoch training. The demonstration that DETM 'captures' discourse transitions is therefore a paraphrase of the fitted word probabilities from which the transitions were inferred, not an independent prediction. The abstract's effectiveness claim rests on this same self-referential evidence.

full rationale

This paper is a descriptive topic-modeling application: most of its narrative is a direct reading of DETM's fitted topic-word distributions, which is the normal output of such a model and not circular by itself. However, two load-bearing choices create genuine partial circularity. First, the paper selects topic 2 post hoc using a predefined set of CCS-related keywords, then reports that topic 2 exhibits a CCS focus; the presence of CCS vocabulary in the chosen topic is true by construction, and the 'technical mitigation strategies' period is built from the selection keywords. Second, the claim that DETM is a 'scalable and effective tool' is supported only by the same fitted beta values that generated the observed trends; no baseline, stability analysis, or external validation is provided, so the effectiveness demonstration reduces to a restatement of the model output. The analysis is not wholly vacuous because the temporal trends align with externally known milestones (Kyoto Protocol, Paris Agreement, Santiago Network), providing some independent grounding. There is no self-citation chain or imported uniqueness theorem; the circularity is localized to topic selection and the self-referential validation of DETM. Score 4 reflects partial circularity in the central claim rather than a fully forced derivation.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The paper does not introduce new theoretical entities, but the analysis depends on several domain assumptions about the corpus and the model. The free parameters are largely hyperparameters chosen by the authors without systematic validation, and the post hoc selection of topic 2 shapes the main findings. The central claim rests on these choices, making them load-bearing.

free parameters (7)
  • num_topics = 5
    Number of topics set experimentally without reported validation or sensitivity analysis (Section 3.2).
  • min_df = 100
    Minimum document frequency threshold for vocabulary retention, set to balance vocabulary size and informativeness (Section 3.1).
  • t_hidden_size = 800
    Hidden state size for the topic proportion network, configured experimentally (Section 3.2).
  • rho_size = 300
    Embedding size, aligned with pre-trained word embeddings (Section 3.2).
  • train_epochs = 5
    Number of training epochs deemed sufficient based on validation loss (Section 3.2).
  • visualization_interval = 6 years
    Six-year intervals chosen to capture long-term trends, which shapes the interpretation of temporal evolution (Section 3.3).
  • topic_selection = topic 2
    Topic 2 selected based on relevance to carbon capture and storage, a post hoc choice that focuses the main narrative (Section 3.3).
assumptions (4)
  • domain assumption UNFCCC decision documents, split into paragraphs, are a representative sample of global climate policy discourse.
    The paper treats the corpus as capturing the evolution of policy language (Sections 1 and 3.1).
  • domain assumption Paragraph-level granularity preserves thematic coherence better than document-level granularity.
    Stated in Section 3.1 without empirical justification.
  • domain assumption DETM's generative assumptions hold: topics evolve smoothly over time via a recurrent neural network, and word embeddings capture semantic relatedness.
    The paper relies on DETM's assumptions and cites the original method (Section 3.2, reference [8]).
  • ad hoc to paper A set of five topics is sufficient to capture the discourse structure of the corpus.
    The number of topics was set experimentally with no sensitivity analysis or diagnostic to support the choice (Section 3.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Temporal Analysis of Climate Policy Discourse: Insights from Dynamic Embedded Topic Modeling." pith.science (2026). https://pith.science/paper/WHGPUCUX

@misc{pith2026250706435,
  author       = {Pith},
  title        = {Pith review of: Temporal Analysis of Climate Policy Discourse: Insights from Dynamic Embedded Topic Modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WHGPUCUX}},
  note         = {Machine review of arXiv:2507.06435}
}
read the original abstract

Understanding how policy language evolves over time is critical for assessing global responses to complex challenges such as climate change. Temporal analysis helps stakeholders, including policymakers and researchers, to evaluate past priorities, identify emerging themes, design governance strategies, and develop mitigation measures. Traditional approaches, such as manual thematic coding, are time-consuming and limited in capturing the complex, interconnected nature of global policy discourse. With the increasing relevance of unsupervised machine learning, these limitations can be addressed, particularly under high-volume, complex, and high-dimensional data conditions. In this work, we explore a novel approach that applies the dynamic embedded topic model (DETM) to analyze the evolution of global climate policy discourse. A probabilistic model designed to capture the temporal dynamics of topics over time. We collected a corpus of United Nations Framework Convention on Climate Change (UNFCCC) policy decisions from 1995 to 2023, excluding 2020 due to the postponement of COP26 as a result of the COVID-19 pandemic. The model reveals shifts from early emphases on greenhouse gases and international conventions to recent focuses on implementation, technical collaboration, capacity building, finance, and global agreements. Section 3 presents the modeling pipeline, including preprocessing, model training, and visualization of temporal word distributions. Our results show that DETM is a scalable and effective tool for analyzing the evolution of global policy discourse. Section 4 discusses the implications of these findings and we concluded with future directions and refinements to extend this approach to other policy domains.

Figures

Figures reproduced from arXiv: 2507.06435 by the authors.

Figure 1
Figure 1. Flow diagram of study Preprocessing was performed to prepare the texts for DETM training. Using the DETM [8] prepro￾cessing script, we applied standard NLP techniques. We split the texts into paragraphs to ensure finer granularity, as paragraph-level analysis better captures thematic coherence. We filtered out stopwords, punctuation, and low-frequency terms. To balance vocabulary size and informativeness, we set a m… view at source ↗
Figure 2
Figure 2. Temporal evolution of topic 2 ("Climate Policy Mechanisms and Mitigation Strategies") [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Evolution of query words in topic 2, showing the probability trends of key terms [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Temporal evolution of topic 1 ("International Climate Agreements and Frameworks") [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Temporal evolution of topic 3 ("Conference and Implementation Processes") [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Temporal evolution of topic 4 ("Technological and Commitment Foundations") [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Temporal evolution of topic 5 ("Emission Management and Reporting") [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 21 canonical work pages

  1. [26]

    Saving the United Nations Framework Convention on Climate Change from itself: Complementary and alternative approaches,

    A. Antypas, "Saving the United Nations Framework Convention on Climate Change from itself: Complementary and alternative approaches,"Ekistics and the New Habitat: The Problems and Science of Human Settlements, vol. 82, no. 3, pp. 3-8, Jun. 2024, doi:10.53910/26531313- E2022823668

    work page doi:10.53910/26531313- 2024

  2. [1]

    Pörtneret al., Eds., Cambridge, U.K.: Cambridge Univ

    Intergovernmental Panel on Climate Change, “Africa,” inClimate Change 2022 – Impacts, Adaptation and Vulnerability: Working Group II Contribution to the Sixth Assessment Report of 6 the Intergovernmental Panel on Climate Change, H.-O. Pörtneret al., Eds., Cambridge, U.K.: Cambridge Univ. Press, 2023, pp. 1285–1456, doi:10.1017/9781009325844

    doi:10.1017/9781009325844 2022

  3. [2]

    The UNFCCC and the Paris Agreement,

    Y . Yamineva and N. Löther, “The UNFCCC and the Paris Agreement,” inElgar Encyclopedia of Climate Policy, D. J. Fiorino, T. A. Eisenstadt, and M. K. Ahluwalia, Eds., Cheltenham, U.K.: Edward Elgar Publishing, 2024, pp. 249-252, doi:10.4337/9781802209204.ch47

    work page doi:10.4337/9781802209204.ch47 2024

  4. [3]

    Non-state actors in global climate governance: From copen- hagen to paris and beyond,

    K. Bäckstrandet al., “Non-state actors in global climate governance: From copen- hagen to paris and beyond,”Environ. Polit., vol. 26, no. 4, pp. 561-579, May 2017, doi:10.1080/09644016.2017.1327485

    work page arXiv 2017

  5. [4]

    Global patterns of national climate policies: Analyzing 171 country portfolios on climate policy integration,

    N. M. Schmidt and A. Fleig, “Global patterns of national climate policies: Analyzing 171 country portfolios on climate policy integration,”Environ. Sci. Policy, vol. 84, pp. 177–185, Jun. 2018, doi:10.1016/j.envsci.2018.03.003

    work page doi:10.1016/j.envsci.2018.03.003 2018

  6. [5]

    Diverse climate actors show limited coordination in a large-scale text analysis of strategy documents,

    A. Hsu and R. Rauber, “Diverse climate actors show limited coordination in a large-scale text analysis of strategy documents,”Commun. Earth Environ., vol. 2, no. 30, pp. 1-10, Feb. 2021, doi:10.1038/s43247-021-00098-7

    work page doi:10.1038/s43247-021-00098-7 2021

  7. [6]

    Cooperation and discord in global climate policy,

    R. O. Keohane and D. G. Victor, “Cooperation and discord in global climate policy,”Nature Clim. Change, vol. 6, no. 6, pp. 570-575, May 2016, doi:10.1038/nclimate2937

    doi:10.1038/nclimate2937 2016

  8. [7]

    Probabilistic topic models,

    D. M. Blei, “Probabilistic topic models,”Commun. ACM, vol. 55, no. 4, pp. 77-84, Apr. 2012, doi:10.1145/2133806.2133826

    arXiv 2012

Show all 29 references
  1. [8]

    The dynamic embedded topic model,

    A. B. Dieng, F. J. R. Ruiz, and D. M. Blei, “The dynamic embedded topic model,”arXiv preprint arXiv:1907.05545, 2019

    1907 arXiv

  2. [9]

    Putting it all together,

    D. G. Victor, “Putting it all together,” inGlobal Warming Gridlock: Creating More Effective Strategies for Protecting the Planet. Cambridge, U.K.: Cambridge Univ. Press, 2011, pp. 201–202, doi:10.1017/CBO9780511975714

    2011 doi

  3. [10]

    The effect of the kyoto protocol on carbon emissions,

    R. Aichele and G. Felbermayr, “The effect of the kyoto protocol on carbon emissions,”J. Policy Anal. Manage., vol. 32, no. 4, pp. 731-757, Aug. 2013, doi:10.1002/pam.21720

    2013 doi

  4. [11]

    Institutional and environmental effectiveness: Will the Paris Agreement work?

    R. Dimitrovet al., “Institutional and environmental effectiveness: Will the Paris Agreement work?”WIREs Clim. Change, vol. 10, no. 4, pp. 575-583, Aug. 2019, doi:10.1002/wcc.583

    2019 doi

  5. [12]

    Sustainable finance and implementing the Paris agreement: The key role of jurisdictional legal and regulatory architecture,

    M. Bowman, “Sustainable finance and implementing the Paris agreement: The key role of jurisdictional legal and regulatory architecture,” inSustainable Finance and Climate Change, R. Smits, Ed., Cheltenham, U.K.: Edward Elgar Publishing, 2024, pp. 19-33, doi:10.4337/9781800377288.00012

    2024

  6. [13]

    A. Ullahet al., “Transition towards climate-resilient low-carbon and net-zero future pathways: Investigating the influence of post-Paris Agreement and G-7 climate policies on climate actions in G-20 using quasi-natural experiments,”Sustain. Dev., vol. 33, no. 3, pp. 1-15, Jun....

    2025 doi

  7. [14]

    Topic modeling on historical newspapers,

    T. I. Yang, A. Torget, and R. Mihalcea, “Topic modeling on historical newspapers,” inProc. 5th ACL-HLT Workshop Lang. Technol. Cult. Heritage, Soc. Sci., Humanit., Oregon, USA, Jun. 2011, pp. 96-104. [Online]. Available: https://aclanthology.org/W11-1513.pdf

    2011

  8. [15]

    Latent dirichlet allocation,

    D. M. Blei, A. Y . Ng, and M. I. Jordan, “Latent dirichlet allocation,”J. Mach. Learn. Res., vol. 3, pp. 993-1022, Jan. 2003. [Online]. Available: http://jmlr.csail.mit.edu/papers/v3/blei03a.html

    2003

  9. [16]

    Corporate funding and ideological polarization about climate change,

    J. Farrell, “Corporate funding and ideological polarization about climate change,” in2016 Proc. Natl. Acad. Sci. USA, Jan. 2016, pp. 92-97, doi:10.1073/pnas.1509433112

    2016 doi

  10. [17]

    The spatial distribution of republican and democratic climate opin- ions at state and local scales,

    M. Mildenbergeret al., “The spatial distribution of republican and democratic climate opin- ions at state and local scales,”Climatic Change, vol. 145, no. 3-4, pp. 539-548, Dec. 2017, doi:10.1007/s10584-017-2103-0

    2017 doi

  11. [18]

    Dynamic topic models,

    D. M. Blei and J. D. Lafferty, “Dynamic topic models,” inProc. 23rd Int. Conf. Mach. Learn. (ICML), Pennsylvania, USA, Jun. 2006, pp. 113-120, doi:10.1145/1143844.1143859. 7

    2006

  12. [19]

    Topic detection and tracking for conversational content by using conceptual dynamic latent Dirichlet allocation,

    J.-F. Yeh, Y .-S. Tan, and C.-H. Lee, “Topic detection and tracking for conversational content by using conceptual dynamic latent Dirichlet allocation,”Neurocomputing, vol. 216, pp. 310-318, Dec. 2016, doi:10.1016/j.neucom.2016.08.017

    2016 doi

  13. [20]

    DTM-based analysis of hot topics and evolution of China’s energy policy,

    Z. Wang, R. Zhou, and Y . Wang, “DTM-based analysis of hot topics and evolution of China’s energy policy,”Sustainability, vol. 16, no. 19, pp. 1-21, Oct. 2024, doi:10.3390/su16198293

    2024 doi

  14. [21]

    Topic modeling in embedding spaces,

    A. B. Dieng, F. J. R. Ruiz, and D. M. Blei, “Topic modeling in embedding spaces,”Trans. Assoc. Comput. Linguist., vol. 8, pp. 439–453, Jul. 2020, doi:10.1162/tacl_a_00325

    2020 doi

  15. [22]

    Media attention for climate change around the world: A comparative analysis of newspaper coverage in 27 countries,

    A. Schmidt, A. Ivanova, and M. S. Schäfer, “Media attention for climate change around the world: A comparative analysis of newspaper coverage in 27 countries,”Glob. Environ. Change, vol. 23, no. 5, pp. 1233-1248, Oct. 2013, doi:10.1016/j.gloenvcha.2013.07.020

    2013 doi

  16. [23]

    Sentiment analysis of texts from social networks based on machine learning methods for monitoring public sentiment,

    A. T. Nurlanuly, “Sentiment analysis of texts from social networks based on machine learning methods for monitoring public sentiment,”arXiv preprint arXiv:2502.17143, 2025

    2025 arXiv

  17. [24]

    Critical climate machine: A visual and musical exploration of climate misinformation through machine learning,

    G. Robillard and J. Nika, “Critical climate machine: A visual and musical exploration of climate misinformation through machine learning,”ACM Comput. Graph. Interact. Tech., vol. 7, no. 4, pp. 1-11, Jul. 2024, doi:10.1145/3664215

    2024 doi

  18. [25]

    Global futures of energy, climate, and policy: Qualitative and quantitative foresight towards 2055,

    D. Ansari, F. Holz, and H. B. Tosun, "Global futures of energy, climate, and policy: Qualitative and quantitative foresight towards 2055,"DIW Berlin Discussion Paper, no. 1782, Jan. 2019, doi:10.2139/ssrn.3324127

    2019 doi

  19. [27]

    BERTopic: Neural topic modeling with a class-based TF-IDF procedure,

    M. Grootendorst, “BERTopic: Neural topic modeling with a class-based TF-IDF procedure,” arXiv preprint arXiv:2203.05794, 2022

    2022 arXiv

  20. [28]

    A critical appraisal of the international and national legal framework on climate change,

    H. S. Samaradiwakara, “A critical appraisal of the international and national legal framework on climate change,” inProc. 27th Int. F orestry Environ. Symp., Sri Lanka, Jan. 27-28, 2023, vol. 27, pp. 98–106, doi:10.31357/fesympo.v27.7101

    2023 doi

  21. [29]

    Gender and climate policy: A discursive institutional analysis of Ethiopia’s climate resilient strategy,

    A. A. Mersha and F. van Laerhoven, “Gender and climate policy: A discursive institutional analysis of Ethiopia’s climate resilient strategy,”Reg. Environ. Change, vol. 19, no. 2, pp. 429-440, Feb. 2019, doi:10.1007/s10113-018-1413-8. 8 Appendix This section provides additional...

    2019 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.