From Newswire to Nexus: Using text-based actor embeddings and transformer networks to forecast conflict dynamics

Mihai Croicu; Simon Polichinel von der Maase

REVIEW 4 major objections 4 minor 1 cited by

From Newswire to Nexus: Using text-based actor embeddings and transformer networks to forecast conflict dynamics

T0 review · 4 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Newswire text alone can forecast when armed conflicts escalate or de-escalate up to six months ahead.

desk verdict The paper has a genuinely new text-to-forecast pipeline, but the evaluation target is leaky: the Gaussian process that defines escalation states is fitted on data through the end of the test window, so the reported skill is measured against an ex-post smoothed target, not a real-time observable. read the letter →

arxiv 2501.03928 v1 pith:HTSVOOEQ submitted 2025-01-07 cs.CY cs.CLcs.LG

Mihai Croicu , Simon Polichinel von der Maase This is my paper · ORCID

classification cs.CYcs.CLcs.LG

keywords conflictforecastingtext-basedpredictionactordyadsescalationandde-escalationtransformerfine-tuningnewswirecorporaGaussianprocesstargetearlywarningsystems

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

Armed-conflict forecasting has mostly been built on lagged fatality counts and spatial history, which predict the next fatality level better than they predict changes. This paper tries to establish that the raw newswire text behind conflict event data carries an earlier, richer signal about actor-level dynamics: whether a fighting pair is escalating, de-escalating, plateauing, or at peace. It constructs a large corpus by matching manually coded conflict events to news articles, uses a fine-tuned language model to label the rest of the corpus by actor dyad, and builds monthly digests of violent and contextual articles for each dyad. Fine-tuned encoder-based transformers trained on these digests predict the discretized slope of the smoothed fatality trend, and the paper reports that they beat a strong historical baseline substantially for nowcasts and one-month-ahead forecasts, with skill decaying by six months. If right, text-trained models could generate early warning of escalation at the actor level without waiting for manual event-data curation at prediction time.

What carries the argument

The carrying machinery is a pipeline ending in a four-state momentum target plus a fine-tuned transformer. Momentum is defined as the discretized first derivative of a hierarchical Gaussian process fitted to each dyad's monthly log fatalities, with a Matérn 3/2 kernel and a shared prior on the length scale; a threshold of $\pm 0.25$ on the derivative splits escalation, plateau, and de-escalation. From the text side, a lightweight BERT-class model is fine-tuned on manually labeled event articles and used to assign every remaining article to one of 25 well-reported actor dyads, filtering out irrelevant news. A topic-model and retrieval step augments each violent event's snippet with nearby non-violent context snippets, and the snippets are concatenated into long dyad-month digests. The forecasting model is a fine-tuned DeBERTa-v3-large encoder with a softmax classification head, trained by step-shifting: predicting the target shifted 0, 1, 3, or 6 months forward, with a 7-billion-parameter decoder model, Mistral 7B, included for comparison.

What would settle it

Re-run the same models against a target computed causally, for example a Gaussian process or moving average updated only with fatalities available at the forecast month, and check whether the DeBERTa models still beat the bootstrapped baseline by the reported margins at the one-month and three-month horizons; if the advantage largely vanishes, the headline skill was an artifact of the ex-post smoothed target.

Watch

Extended reading notes

Core claim

The central claim is that the textual record of a conflict dyad's month contains usable predictive signal about the direction its violence is heading, and that a model can extract that signal directly from text. The paper's target is not the level of fatalities but the momentum: a Gaussian process is fitted to monthly log-fatality series for each dyad, and the first derivative of that smoothed trend is discretized into four states: peace, escalation, plateau, and de-escalation. Against a baseline that bootstraps the previous year's observed state distribution, two fine-tuned DeBERTa-based models, one reading only violent-event snippets and one reading events augmented with retrieved contextual articles, exceed the baseline on average precision, AUROC, and F1 for nowcasting and one-month-ahead forecasting, remain above it at three months, and fall to near-baseline at six months. The paper further claims this makes the manual curation step unnecessary at prediction time, and it reports that the much larger decoder-based Mistral model performs poorly, below baseline, a result the authors offer as a caution about decoder-only models.

Load-bearing premise

The load-bearing premise is that a smoothed escalation state computed from a curve fitted on fatality data spanning the whole test period can stand in for the value a forecaster would have wanted to predict in real time, even though that curve incorporates fatalities that occurred after the forecast month.

Editorial extensions

If this is right

Conflict early-warning systems could skip the weeks-long manual event-data annotation step at prediction time and read newswire text directly for escalation signals.
The text-only signal is strongest for escalation and weaker for de-escalation, so users should expect asymmetric skill and weight early-warning products accordingly.
The collapse of skill between three and six months defines a practical forecast horizon for text-based momentum models.
Large decoder-only generative models are not automatically better at this task; smaller encoder-based models can dominate them, so architecture choice matters.
Retrieval-style context augmentation helps at most horizons but not dramatically, suggesting the main predictive gain comes from the violent-event text itself.

Reading between the lines

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

A direct extension, not run in the paper, is to recompute the target causally from only the fatalities known at the forecast month; if the reported margins survive, the claim is robust to the target-leakage concern, and if not, part of the reported skill is an artifact of ex-post smoothing.
The authors' own reading leaves open, and we infer, that much of the signal lives in the violent-event text rather than in the added non-violent context, since the RAG augmentation yields only modest gains.
We infer, beyond the paper's dyad-selection caveat, that the margin over baseline should not be extrapolated to rarely reported or newly emerging conflicts, where the news corpus and the training labels are far thinner.

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.

Desk Editor's Note

read the letter

The core idea here is worth taking seriously: use UCDP's human annotations to back-label a large newswire corpus, build dyad-month digests, and fine-tune DeBERTa (and Mistral) to forecast a momentum target rather than incidence. That is a real departure from topic-proportion news models and from earlier actor-centric work. The pipeline is described carefully, with intermediate evaluations, and the authors are honest that Mistral fails and that the six-month results are essentially at baseline. I also appreciate that they pick a strong bootstrap baseline rather than a weak one.

The problem is the target. Section 3.5 says they fit two GPs—one for training through December 2022 and one for validation through March 2023—and then use the discretized first derivative as the target for both training and validation. The validation labels for a month in the test window are therefore a function of fatalities observed after that month. A GP smoother at time x uses observations on both sides of x, so every test-month label in the Jan 2022–Mar 2023 window is retroactively smoothed with up to three months of future data. That is look-ahead bias in the evaluation. The paper claims this setup prevents leakage, but it only prevents leakage between the two GP fits; it does not make the target causal. The abstract's claim of skill "up to six months" is also contradicted by the paper's own Figure 4, where the DeBERTa models are near baseline at six months. And there is an inconsistency: the declared training split ends December 2021, yet the training GP is fit through December 2022, which if taken literally means the training targets also embed future information.

The consequence is that the reported AP/F1 advantage over the bootstrap baseline measures how well text predicts a retrospectively smoothed state, not how well it predicts what an analyst could have known at forecast time. That is not a minor benchmark quibble; it is the central forecasting claim. What remains is a well-built proof-of-concept that text contains some signal about conflict dynamics, but the headline result as stated is not supported.

My recommendation: send it to review. The pipeline and the questions it asks are substantial enough to deserve referee time, but the revision path is clear—redefine the target causally, using for each month only data available at that month (e.g., a filtered derivative or a one-sided smoother), and re-run the evaluation. If the results survive that, this becomes a solid contribution. If they do not, the paper is still a useful methodological case study, but the forecasting claim has to go.

Referee Report

4 major / 4 minor

Summary. The paper proposes an end-to-end pipeline that forecasts dyad-month conflict dynamics from newswire text. It curates a Factiva corpus using UCDP GED annotations, trains a ConfliBERT-based classifier to assign unlabeled articles to conflict dyads, constructs low- and high-context monthly digests via BERTopic and retrieval-augmented generation, and defines the forecast target as the discretized first derivative of a hierarchical Gaussian process fitted to log fatalities. The authors fine-tune DeBERTa-v3-large and Mistral 7B with step-shifting at nowcasting and 1-, 3-, and 6-month horizons, and compare against a bootstrap baseline. The reported results show DeBERTa outperforming the baseline for nowcasting and one-month-ahead forecasts, attenuating at three months, and near collapse at six months, while Mistral performs poorly.

Significance. If the evaluation were valid, this would be a substantial contribution to conflict forecasting: it moves from incidence/severity targets to escalation dynamics, operates at the actor-dyad level, and demonstrates that text-only inputs can replace manually curated event data at prediction time. The pipeline is transparent and the authors deserve credit for reporting the Mistral failure and the six-month attenuation honestly. However, the central out-of-sample claim is currently undermined by two evaluation problems: the validation target is an ex-post smoothed quantity, and hyperparameter tuning was performed on the test window. Both issues are load-bearing for the headline result and need to be addressed before the forecasting claim can be assessed.

major comments (4)

[Section 3.5, Eq. (4)] The validation target is not observable at forecast time. The paper states: 'To prevent data leakage from the training to the validation partition, we estimate two Gaussian processes – one f^train,dyad trained on data until the end of the training window (December 2022) and one used for validation f^val,dyad until the end of the data (March 2023).' Because a Gaussian process smoother at month x uses observations from both sides of x, every validation label in the January 2022–March 2023 window is a function of fatalities observed after the forecast month, in the worst case as far ahead as March 2023. Thus the reported AP/F1 scores for nowcasting and one-month-ahead forecasts are scores against an ex-post smoothed target, not against anything knowable when the forecast would be issued. This is not a minor benchmark quibble: both the DeBERTa models and the bootstrap baseline are compared on these leaky labels. Please re-estimate the target in a causal manner (e.g., sequential GP fits using only data up to the forecast month, or a one-sided filter) and re-report all metrics, including the baseline, on the causal labels.
[Section 3.6, footnote 34] Hyperparameter tuning for the DeBERTa models was performed on 'a small randomly sampled dataset from the test window.' Since the test window is January 2022–March 2023, using it for model selection means the reported test-set metrics are not strictly out-of-sample; the headline comparison to the baseline is therefore potentially optimistic. Please move all hyperparameter selection to a validation set drawn from the training period, or use a nested cross-validation scheme, and report whether the conclusions change.
[Section 5 and Figure 4] The abstract and conclusion claim that newswire texts 'can be used to effectively predict the momentum of armed conflicts ... over periods of up to six months,' but this is contradicted by the paper's own results in Section 4 and Figure 4, where at six months the DeBERTa models are described as 'just slightly better in AP and F1 than the baseline, and ... at baseline for AUROC.' The supported claim is for nowcasting and one-month-ahead horizons, with attenuation at three months. Please revise the abstract and conclusion to state the horizons for which the evidence actually holds.
[Section 3.1 vs. Section 3.5] The description of the training window is internally inconsistent. Section 3.1 states that the data are divided 'in a training set (until the end of 2021) and a test set (starting January 2022),' whereas Section 3.5 says f^train is trained 'until the end of the training window (December 2022).' If the training digests stop in December 2021 while training targets are derived from a GP fitted through December 2022, then the target attached to the final training months encodes up to twelve months of future information. Please clarify the actual training window and, if needed, recompute training targets with point-in-time information.

minor comments (4)

[Section 3.5, Eq. (3)] The text refers to the 'Matérn 3/2 covariate function'; this should be 'covariance function.' Also, the length-scale parameter ℓ is introduced in Eq. (3) but defined only in the following prose; please define it at first use.
[Figure 4 caption] The caption says 'micro-aggregation where probabilities are involved (ROC, AP),' but the figure also reports F1 micro and F1 weighted. Please clarify how each metric is aggregated and how the bootstrap confidence intervals are computed (e.g., resampling units).
[Section 3.5, Eq. (4)] The target definition depends on the threshold τ=0.25 and on an informative length-scale prior (MAP 122.38) taken from von der Maase (2022b), a working paper by the second author. A sensitivity analysis over τ and the prior, or at least a discussion of their influence, would strengthen confidence in the target construction.
[References] von der Maase (2022b) is cited as the source for the Gaussian-process target definition, but it is a working paper. Please provide a stable reference or an accessible replication appendix, since the evaluation target depends on that method.

Circularity Check

1 steps flagged · score 2.0 of 10

No circular derivation: text inputs are mapped to an external fatality-derived target; only minor self-reliance in the GP target estimator and a separate temporal-leakage concern.

other [Section 2.5 and Section 3.5, Eqs. (2)-(3)]
"We adapt the solution provided by von der Maase (2022b)... For each dyad the estimation is started from a global highly informative prior drawn from von der Maase, 2022b’s findings – a LogNormal centered around the main temporal long-term finding in that paper, the estimated global maximum a posteriori (MAP) long-term ℓ of 122.38."

The four escalation states are defined by discretizing the derivative of a Gaussian process, and the kernel choice plus the informative length-scale prior are justified by the second author's unpublished working paper rather than derived in the present manuscript. This is a self-citation in the target-construction chain. It is not a full circularity: the forecasting models do not receive the GP labels as inputs; they map text digests to these fatality-derived labels, so the central claim does not reduce to the citation. It is flagged as a minor self-reliance issue.

full rationale

The core derivation is self-contained with respect to the circularity concerns. Newswire digests are compiled in Section 3.4, then fine-tuned DeBERTa and Mistral classifiers are trained in Section 3.6 with step-shifted targets; the target S(x) in Eq. (4) comes from a GP fitted to UCDP fatality counts, not from the text inputs. No equation in the paper defines the predicted states as a function of the input texts by construction, and the bootstrap baseline is computed from past fatalities, so the reported AP/F1 advantages are empirical comparisons rather than algebraic identities. The one self-referential element is the GP target estimator: the Matérn kernel and the LogNormal prior with mean length-scale 122.38 are imported from von der Maase (2022b), an unpublished working paper by the second author, and this choice shapes all target labels. That is a minor self-citation, but the text-to-state mapping is still an independent empirical fit. Separately, Section 3.5 states that the validation GP is fit through March 2023, so validation labels for earlier test months are retroactively smoothed using future fatalities; this is a temporal-leakage and target-validity concern, not a circular reduction, and therefore does not raise the circularity score under the stated rules.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claims rest on a small number of hand-set parameters (escalation threshold, assignment probability, digest sizes) and on domain assumptions about newswire structure and the validity of the GP-derived target. No new theoretical entities are introduced.

free parameters (7)

tau (threshold in Eq. 4) = 0.25
Hand-set threshold for discretizing the GP first derivative into escalation, plateau, and de-escalation states; chosen to avoid estimation wobble.
GP length-scale prior (MAP) = 122.38 months
Global maximum a posteriori length scale taken from von der Maase (2022b), used as the mean of the informative prior for each dyad's Gaussian process.
Article-dyad assignment probability threshold = 0.8
Articles with less than 0.8 predicted probability of belonging to one of the 25 dyads are discarded in Section 3.3.
BERTopic minimum topic size = 200 articles
Minimum number of articles per topic when fitting BERTopic per dyad-month, yielding 3 to 21 topics per dyad (Section 3.4).
Context snippet count = 5 (low-context) / 1 per topic (high-context)
Number of nearest articles to topic centroid (low-context) or to each event (high-context RAG) used to build digests (Section 3.4).
Snippet length / token truncation = 256 tokens
Articles truncated to the first 256 tokens based on the assumption that dyad identity appears in head and lead paragraphs (Section 3.3).
Selected dyads = 25
Restriction to the 25 most-reported dyads in June to December 2021, which limits generalizability and changes class balance.

assumptions (6)

domain assumption Head and lead paragraphs of a newswire article contain the identity of the fighting dyad
Invoked in Section 3.3 to justify truncating all articles to 256 tokens.
domain assumption Headline matching between Factiva articles and UCDP GED events correctly assigns dyads
Section 3.2 uses headline matching to pair 54,669 articles with UCDP events; false matches propagate into training labels.
domain assumption The first derivative of a Gaussian process fitted to log-fatalities, discretized with a hand-set threshold, is a valid measure of conflict dynamics
Section 3.5 defines the target S(x) via Eq. (4). The validity of this target is assumed, and the GP fit uses future data, making the target non-causal.
domain assumption The UCDP keyword query {kill*, die*, injur*, dead*, wound*, massacre*} captures the conflict-relevant newswire universe
Section 3.1 uses this query to download the corpus; articles outside these stems, including many non-violent context pieces, are missed.
standard math Standard Gaussian process regression results, including the Matern 3/2 kernel and derivative properties
Used in Section 3.5 without proof.
domain assumption Fine-tuned transformer models generalize from the labeled split to the unlabeled corpus without distribution shift
Section 3.3 applies the ConfliBERT classifier to 2.1 million unmatched articles; the 0.75 to 0.80 recall measured on a random split is assumed to hold on the full corpus.

how reviews work

0 comments

Cite this review

Pith. "Pith review of From Newswire to Nexus: Using text-based actor embeddings and transformer networks to forecast conflict dynamics." pith.science (2026). https://pith.science/paper/HTSVOOEQ

@misc{pith2026250103928,
  author       = {Pith},
  title        = {Pith review of: From Newswire to Nexus: Using text-based actor embeddings and transformer networks to forecast conflict dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HTSVOOEQ}},
  note         = {Machine review of arXiv:2501.03928}
}

read the original abstract

This study advances the field of conflict forecasting by using text-based actor embeddings with transformer models to predict dynamic changes in violent conflict patterns at the actor level. More specifically, we combine newswire texts with structured conflict event data and leverage recent advances in Natural Language Processing (NLP) techniques to forecast escalations and de-escalations among conflicting actors, such as governments, militias, separatist movements, and terrorists. This new approach accurately and promptly captures the inherently volatile patterns of violent conflicts, which existing methods have not been able to achieve. To create this framework, we began by curating and annotating a vast international newswire corpus, leveraging hand-labeled event data from the Uppsala Conflict Data Program. By using this hybrid dataset, our models can incorporate the textual context of news sources along with the precision and detail of structured event data. This combination enables us to make both dynamic and granular predictions about conflict developments. We validate our approach through rigorous back-testing against historical events, demonstrating superior out-of-sample predictive power. We find that our approach is quite effective in identifying and predicting phases of conflict escalation and de-escalation, surpassing the capabilities of traditional models. By focusing on actor interactions, our explicit goal is to provide actionable insights to policymakers, humanitarian organizations, and peacekeeping operations in order to enable targeted and effective intervention strategies.

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

Do Large Language Models Know Conflict? Investigating Parametric vs. Non-Parametric Knowledge of LLMs for Conflict Forecasting
cs.CL 2025-05 conditional novelty 4.0 of 10

LLMs show limited parametric knowledge for conflict forecasting; adding retrieved context from GDELT and ACLED improves GPT-4's predictions modestly but does not help LLaMA-2.

Reference graph

Works this paper leans on

10 extracted references · 6 canonical work pages · cited by 1 Pith paper

[1]

Why attentions may not be interpretable?

Achiam J, Adler S, Agarwal S, et al. (2023) GPT-4 technical report.arXiv preprint arXiv:2303 .08774: URL: https://arxiv.org/abs/2303.08774. Amos M, Pinder T and Young P J (2022) Identifying latent climate signals using sparse hierarchical Gaussian processes. NeurIPS Workshop on Gaussian Pro- cesses, Spatiotemporal Modeling, and Decision-making Systems. Ba...

arXiv 2023
[2]

MIT Press Cambridge, MA. Williams JP, Hermansen GH, Strand H, Clayton G and Nygård HM (2024) Bayesian hidden Markov models for latent variable labeling assignments in conflict research: application to the role ceasefires play in conflict dynamics. The Annals of Applied Statistics 18(3): 2034–2061. Ziems C, Held W, Shaikh O, Chen J, Zhang Z and Yang D (202...

work page 2024
[56]

(2018) Improving lan- guage understanding by generative pre-training

Radford A, Narasimhan K, Salimans T, Sutskever I, et al. (2018) Improving lan- guage understanding by generative pre-training. Open AI Publication: URL: Croicu and von der Maase | From Newswire to Nexus | Paper V 33 https : / / cdn . openai . com / research - covers / language - unsupervised / language_understanding_paper.pdf. Radford BJ (2022) High resol...

work page 2018
[67]

Computational enhancements of HNSW targeted to very large datasets

New York: Peter Lang. Factiva (2022) Annual Content RoundUp – Factiva for ProQuest. https : / / pq - static - content . proquest . com / collateral / media2 / documents / factiva - annualroundup.pdf. Accessed: 2023-03-21. ProQuest. Foster C and Kimia B (2023) “Computational enhancements of HNSW targeted to very large datasets”. In: International Conferenc...

arXiv 2022
[375]

Journal of the European Economic Association 20(6): 2440–2467

Mueller H and Rauh C (2022a) The hard problem of prediction for conflict preven- tion. Journal of the European Economic Association 20(6): 2440–2467. Mueller H and Rauh C (2022b) Using past violence and current news to predict changes in violence. International Interactions 48(4): 579–596. Mueller H, Rauh C and Seimon B (2024) Introducing a global dataset...

work page 2022
[660]

Automated coding of political event data

Randahl D and Vegelius J (2022) Predicting escalating and de-escalating violence in Africa using Markov models. International Interactions 48(4): 597–613. Rød EG, Gåsste T and Hegre H (2024) A review and comparison of conflict early warning systems. International Journal of Forecasting 40(1): 96–112. Schrodt PA and Van Brackle D (2012) “Automated coding o...

work page 2022
[1056]

Transfer learning

Torrey L and Shavlik J (2010) “Transfer learning”. In:Handbook of research on ma- chine learning applications and trends: algorithms, methods, and techniques. IGI global, pp. 242–264. Vaswani A, Shazeer N, Parmar N, et al. (2017) “Attention is all you need”. In: Proceedings of the 31st International Conference on Neural Information Pro- cessing Systems. N...

work page 2010
[2008]

Grootendorst M (2022) BERTopic: Neural topic modeling with a class-based TF-IDF procedure

05756. Grootendorst M (2022) BERTopic: Neural topic modeling with a class-based TF-IDF procedure. arXiv preprint arXiv:2203.05794: Gustafson D (2020) Hunger to violence: Explaining the violent escalation of non- violent demonstrations. Journal of Conflict Resolution 64(6): 1121–1145. Croicu and von der Maase | From Newswire to Nexus | Paper V 31 Häffner S...

arXiv 2022

Show all 10 references

[2106]

ConfliBERT: A pre-trained language model for political conflict and violence

09685. Hu Y, Hosseini M, Skorupa Parolin E, et al. (2022) “ConfliBERT: A pre-trained language model for political conflict and violence”. In: Proceedings of the Association for Computational Linguistics. Association for Computational Lin- guistics. Hu Z, Wang L, Lan Y, et al. ...

2022 arXiv
[2157]

QLoRA: Efficient finetuning of quantized LLMs

Buhaug H and Gleditsch KS (2008) Contagion or confusion? Why conflicts cluster in space. International Studies Quarterly 52(2): 215–233. Caldwell S (2022) Assessing the technical feasibility of conflict prediction for anticipatory action. Centre for HUMDATA, UN OCHA Services. ...

2008

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Why attentions may not be interpretable?

Achiam J, Adler S, Agarwal S, et al. (2023) GPT-4 technical report.arXiv preprint arXiv:2303 .08774: URL: https://arxiv.org/abs/2303.08774. Amos M, Pinder T and Young P J (2022) Identifying latent climate signals using sparse hierarchical Gaussian processes. NeurIPS Workshop on Gaussian Pro- cesses, Spatiotemporal Modeling, and Decision-making Systems. Ba...

arXiv 2023

[2] [2]

MIT Press Cambridge, MA. Williams JP, Hermansen GH, Strand H, Clayton G and Nygård HM (2024) Bayesian hidden Markov models for latent variable labeling assignments in conflict research: application to the role ceasefires play in conflict dynamics. The Annals of Applied Statistics 18(3): 2034–2061. Ziems C, Held W, Shaikh O, Chen J, Zhang Z and Yang D (202...

work page 2024

[3] [56]

(2018) Improving lan- guage understanding by generative pre-training

Radford A, Narasimhan K, Salimans T, Sutskever I, et al. (2018) Improving lan- guage understanding by generative pre-training. Open AI Publication: URL: Croicu and von der Maase | From Newswire to Nexus | Paper V 33 https : / / cdn . openai . com / research - covers / language - unsupervised / language_understanding_paper.pdf. Radford BJ (2022) High resol...

work page 2018

[4] [67]

Computational enhancements of HNSW targeted to very large datasets

New York: Peter Lang. Factiva (2022) Annual Content RoundUp – Factiva for ProQuest. https : / / pq - static - content . proquest . com / collateral / media2 / documents / factiva - annualroundup.pdf. Accessed: 2023-03-21. ProQuest. Foster C and Kimia B (2023) “Computational enhancements of HNSW targeted to very large datasets”. In: International Conferenc...

arXiv 2022

[5] [375]

Journal of the European Economic Association 20(6): 2440–2467

Mueller H and Rauh C (2022a) The hard problem of prediction for conflict preven- tion. Journal of the European Economic Association 20(6): 2440–2467. Mueller H and Rauh C (2022b) Using past violence and current news to predict changes in violence. International Interactions 48(4): 579–596. Mueller H, Rauh C and Seimon B (2024) Introducing a global dataset...

work page 2022

[6] [660]

Automated coding of political event data

Randahl D and Vegelius J (2022) Predicting escalating and de-escalating violence in Africa using Markov models. International Interactions 48(4): 597–613. Rød EG, Gåsste T and Hegre H (2024) A review and comparison of conflict early warning systems. International Journal of Forecasting 40(1): 96–112. Schrodt PA and Van Brackle D (2012) “Automated coding o...

work page 2022

[7] [1056]

Transfer learning

Torrey L and Shavlik J (2010) “Transfer learning”. In:Handbook of research on ma- chine learning applications and trends: algorithms, methods, and techniques. IGI global, pp. 242–264. Vaswani A, Shazeer N, Parmar N, et al. (2017) “Attention is all you need”. In: Proceedings of the 31st International Conference on Neural Information Pro- cessing Systems. N...

work page 2010

[8] [2008]

Grootendorst M (2022) BERTopic: Neural topic modeling with a class-based TF-IDF procedure

05756. Grootendorst M (2022) BERTopic: Neural topic modeling with a class-based TF-IDF procedure. arXiv preprint arXiv:2203.05794: Gustafson D (2020) Hunger to violence: Explaining the violent escalation of non- violent demonstrations. Journal of Conflict Resolution 64(6): 1121–1145. Croicu and von der Maase | From Newswire to Nexus | Paper V 31 Häffner S...

arXiv 2022