arxiv: 2605.01528 · v1 · submitted 2026-05-02 · 💻 cs.SI

Recognition: unknown

Shifting Patterns of Extremist Discourse on Facebook: Analyzing Trends and Developments During the Israel-Hamas Conflict

Rr. Nefriana , Muheng Yan , Ahmad Diab , Wanhao Yu , Deborah L. Wheeler , Andrew Miller , Rebecca Hwa , Yu-Ru Lin

Authors on Pith no claims yet

Pith reviewed 2026-05-10 15:41 UTC · model grok-4.3

classification 💻 cs.SI

keywords extremist discourseFacebook groupsIsrael-Hamas conflictsentiment trendstopic shiftsonline hatesocial media analysis

0 comments

The pith

Extremist Facebook groups adjusted activity levels, negativity, and topics in response to specific Israel-Hamas conflict events.

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

This paper tracks engagement, sentiment, and topics in Facebook groups divided into anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both categories over a one-year period that includes the recent war. It tests whether these measures shifted in line with five major conflict milestones. Activity share dropped in mixed groups and rose in one-sided groups at the war's start, then reversed after Israeli forces withdrew from Khan Yunis. Negative posts increased and neutral posts decreased across every category once fighting began. Topics in anti-Palestine groups moved toward activism and protest while anti-Israel groups turned toward religious content around the same period.

Core claim

Shifts in trends of extremist discourse correspond with key events in the Israel-Hamas conflict, showing varying patterns across group categories. Activity proportion decreased in anti-both groups and increased in the two one-sided hate groups at the conflict's onset, reversing after the Israeli troop withdrawal from Khan Yunis. Negative content proportion rose and neutral content fell in all three categories during the conflict. Anti-Palestine/Muslim groups shifted from religious to social media activism and political/protest topics near the war's start, while anti-Israel/Semitic groups moved from political/protest to religious topics weeks before the war.

What carries the argument

Temporal mapping of activity proportions, sentiment distributions, and topic categories in three predefined Facebook group types against a timeline of five conflict events.

Load-bearing premise

The Facebook groups were correctly and stably placed into the anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both categories with no major misclassification or selection bias.

What would settle it

If activity proportions, negative sentiment share, and topic distributions showed no measurable change or reversal around the dates of the five major events including the conflict onset and the withdrawal from Khan Yunis.

Figures

Figures reproduced from arXiv: 2605.01528 by Ahmad Diab, Andrew Miller, Deborah L. Wheeler, Muheng Yan, Rebecca Hwa, Rr. Nefriana, Wanhao Yu, Yu-Ru Lin.

**Figure 2.** Figure 2: The percentages of posts with their corresponding sentiment on the Facebook groups [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 3.** Figure 3: The distribution of topics on the Facebook groups over time (from July 1st, 2023, to June [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

read the original abstract

This short paper explores trends in extremist Facebook data from July 2023 to June 2024. We examined engagement, sentiment, and topics within Facebook groups categorized as anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both, mapping these trends against five major events related to the recent Israel-Hamas conflict. Our findings support the hypothesis that shifts in trends correspond with these key events, showing varying patterns across different group categories. We observed decreased activity proportion in anti-both groups and increased activity proportion in the two one-sided hate groups at the conflict's onset. This pattern reversed after the Israeli troop withdrawal from Khan Yunis, Gaza. During the conflict, negative content proportion surged, and neutral content proportion fell in all the three group categories. Anti-Palestine/Muslim groups' discourses shifted from religious to social media activism and political/protest around the time the war began, while anti-Israel/Semitic groups moved from political/protest to religious topics a couple of weeks before the war.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper reports some event-aligned shifts in activity and sentiment across three categories of Facebook extremist groups during the Israel-Hamas conflict, but the complete lack of detail on how those groups were labeled undercuts the claims.

read the letter

The main point is that this paper tracks changes in activity levels, sentiment, and discussion topics across three categories of extremist Facebook groups during the Israel-Hamas conflict and links those changes to specific events like the war's start and troop movements. They report that at the conflict's beginning, activity proportion fell in anti-both groups while rising in the one-sided hate groups, then reversed after the Israeli withdrawal from Khan Yunis. Negative content increased and neutral content decreased in all categories during the war. Topics in anti-Palestine groups shifted toward activism and protest, while anti-Israel groups moved toward religious topics. This supplies some fresh observational data points for this particular conflict window. The mapping to dated events is straightforward and parameter-free, which is a plus. The real weakness is in the group categorization. The abstract states that groups were categorized into anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both, but gives no information on the process used. Without details on keywords, coding rules, or validation, it's impossible to judge whether the observed reversals and topic changes reflect real shifts or artifacts of how the groups were chosen or labeled. Sample sizes and any tests for significance are also missing from the summary. The paper applies standard sentiment and topic modeling tools to new data, so the methods are not novel. The value is in the timing and the specific patterns reported. This work is aimed at researchers studying online extremism and platform responses to geopolitical events. Someone looking for case-study evidence on how hate groups react to news events might find the numbers useful as a starting point. I would recommend sending it for peer review once the authors add a clear methods section on data collection, group selection, and labeling validation. The core idea of event-aligned trend analysis is reasonable, but the current version lacks the transparency needed for reliable interpretation.

Referee Report

2 major / 1 minor

Summary. The paper claims to have analyzed engagement, sentiment, and topic trends in Facebook groups categorized into anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both from July 2023 to June 2024. It maps these trends to five major events in the Israel-Hamas conflict and reports that activity proportions decreased in anti-both groups and increased in one-sided groups at the conflict onset, reversing after the Israeli withdrawal from Khan Yunis; negative content surged and neutral fell across categories; and topics shifted from religious to activism in anti-Palestine groups and from political to religious in anti-Israel groups around the war start.

Significance. If the results hold, this work offers timely observational insights into how extremist online discourse on social media adapts to geopolitical conflicts, potentially aiding in the development of monitoring tools for radicalization. The analysis is strengthened by its direct, parameter-free correlation of trends with independently dated external events, without reliance on fitted models or circular definitions.

major comments (2)

[Abstract] The abstract states that Facebook groups were 'categorized as' anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both but provides no information on the categorization method (e.g., keyword rules, manual coding protocol, inter-rater reliability, or temporal stability checks). This is load-bearing because all claims about activity proportion reversals after the Khan Yunis withdrawal, sentiment surges, and topic shifts (religious to activism or vice versa) depend entirely on the accuracy and stability of these labels; without validation, the patterns could reflect misclassification or selection bias rather than genuine event-driven changes.
[Results] No details are given on sample sizes, data collection procedures, or statistical significance tests for the reported proportion changes (e.g., decreased activity in anti-both groups at conflict onset). This omission prevents assessment of whether the observed shifts exceed what would be expected from sampling variability or unstated filters, directly affecting the reliability of the central hypothesis that trends correspond with the five key events.

minor comments (1)

[Abstract] The abstract repeats phrasing such as 'the three group categories' and 'all the three group categories,' which could be streamlined for clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback, which identifies key areas where additional methodological transparency will strengthen the manuscript. We address each major comment below and will incorporate revisions to improve clarity and rigor.

read point-by-point responses

Referee: [Abstract] The abstract states that Facebook groups were 'categorized as' anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both but provides no information on the categorization method (e.g., keyword rules, manual coding protocol, inter-rater reliability, or temporal stability checks). This is load-bearing because all claims about activity proportion reversals after the Khan Yunis withdrawal, sentiment surges, and topic shifts (religious to activism or vice versa) depend entirely on the accuracy and stability of these labels; without validation, the patterns could reflect misclassification or selection bias rather than genuine event-driven changes.

Authors: We agree that the categorization method is central to the validity of the findings and requires explicit documentation. In the revised manuscript, we will add a dedicated Methods subsection detailing the categorization protocol. This will include the specific keyword rules applied for initial group labeling, the manual coding procedure (including the number of coders and any resolution process for disagreements), inter-rater reliability statistics such as Cohen's kappa, and temporal stability checks via re-sampling of groups across the study period. These additions will allow readers to assess potential biases and confirm that the observed shifts align with genuine discourse changes rather than labeling artifacts. revision: yes
Referee: [Results] No details are given on sample sizes, data collection procedures, or statistical significance tests for the reported proportion changes (e.g., decreased activity in anti-both groups at conflict onset). This omission prevents assessment of whether the observed shifts exceed what would be expected from sampling variability or unstated filters, directly affecting the reliability of the central hypothesis that trends correspond with the five key events.

Authors: We concur that reporting sample sizes, data collection details, and statistical tests is essential for evaluating the robustness of the proportion changes. The revised manuscript will explicitly state the data collection procedures (including the platform access method and any filtering criteria applied), report sample sizes such as the number of groups and total posts per category across the July 2023–June 2024 period, and include statistical significance tests (e.g., chi-squared tests for differences in activity proportions at event time points) with associated p-values. This will demonstrate whether the reported reversals and surges exceed sampling variability and support the event-driven hypothesis more rigorously. revision: yes

Circularity Check

0 steps flagged

No circularity: direct empirical mapping of measurements to external events

full rationale

The paper performs straightforward observational analysis: groups are categorized into three types, then activity proportions, sentiment distributions, and topic shifts are measured over time and aligned with independently dated external conflict events. No equations, model fits, predictions, or derivations appear in the abstract or described methods; the central claims reduce only to raw counts and proportions computed from the data, with no self-referential definitions, fitted inputs renamed as predictions, or load-bearing self-citations. The analysis is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central observations rest on the assumption that groups can be reliably partitioned into the three hate categories and that the selected events are the relevant turning points; these are domain assumptions without independent validation or falsifiable tests described in the abstract.

axioms (2)

domain assumption Facebook groups can be accurately and consistently categorized into anti-Israel/Semitic, anti-Palestine/Muslim, and anti-both based on their content.
All trend comparisons are performed separately within these categories.
domain assumption The five major events are the appropriate reference points for detecting shifts in discourse.
Trends are explicitly mapped against these events to support the hypothesis.

pith-pipeline@v0.9.0 · 5503 in / 1573 out tokens · 38565 ms · 2026-05-10T15:41:18.590439+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

16 extracted references · 1 canonical work pages · 1 internal anchor

[1]

International Journal of Communication10, 27 (2016)

Ben-David, A., Fernández, A.M.: Hate speech and covert discrimination on social media: Monitoring the facebook pages of extreme-right political parties in spain. International Journal of Communication10, 27 (2016)

2016
[2]

Retrieved October7, 2020 (2017)

Berecz, T., Devinat, C.: Relevance of cyber hate in europe and current topics that shape online hate speech. Retrieved October7, 2020 (2017)

2020
[3]

In: Pacific-Asia conference on knowledge discovery and data mining

Campello, R.J., Moulavi, D., Sander, J.: Density-based clustering based on hierarchical density estimates. In: Pacific-Asia conference on knowledge discovery and data mining. pp. 160–172. Springer (2013)

2013
[4]

systematic review

Castaño-Pulgarín, S.A., Suárez-Betancur, N., Vega, L.M.T., López, H.M.H.: Internet, social media and online hate speech. systematic review. Aggression and violent behavior58, 101608 (2021)

2021
[5]

In: 2019 International Conference on Advances in Computing and Communication Engineering (ICACCE)

Chetty, N., Alathur, S.: Trigger event and hate content: Insights from twitter analytics. In: 2019 International Conference on Advances in Computing and Communication Engineering (ICACCE). pp. 1–5. IEEE (2019)

2019
[6]

CNN: April 7, 2024 - israel-hamas war (2024),https://www.cnn.com/middleeast/ live-news/israel-hamas-war-gaza-news-04-07-24/index.html, accessed: 2024-07-19

2024
[7]

Unesco Publishing (2015)

Gagliardone, I., Gal, D., Alves, T., Martinez, G.: Countering online hate speech. Unesco Publishing (2015)

2015
[8]

Cambridge University Press (2014)

Gelvin, J.L.: The Israel-Palestine conflict: One hundred years of war. Cambridge University Press (2014)

2014
[9]

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

Grootendorst, M.: Bertopic: Neural topic modeling with a class-based tf-idf procedure. arXiv preprint arXiv:2203.05794 (2022)

work page internal anchor Pith review arXiv 2022
[10]

New Media & Society19(9), 1440–1457 (2017)

Horsti, K.: Digital islamophobia: The swedish woman as a figure of pure and dangerous whiteness. New Media & Society19(9), 1440–1457 (2017)

2017
[11]

Die Welt des Islams 46(3), 243–276 (2006)

Krämer, G.: Anti-semitism in the muslim world: A critical review. Die Welt des Islams 46(3), 243–276 (2006)

2006
[12]

New media & society6(2), 259–283 (2004)

Papacharissi, Z.: Democracy online: Civility, politeness, and the democratic potential of online political discussion groups. New media & society6(2), 259–283 (2004)

2004
[13]

Open Library of Humanities5(1) (2019)

Poole, E., Giraud, E., de Quincey, E.: Contesting# stopislam: The dynamics of a counter- narrative against right-wing populism. Open Library of Humanities5(1) (2019)

2019
[14]

The Jerusalem Post (2023),https://www.jpost.com/israel-hamas-war/article-778165, ac- cessed: 2024-07-19

Post, T.J.: Idf launches major attack on rafah, philadelphi corridor in gaza’s south. The Jerusalem Post (2023),https://www.jpost.com/israel-hamas-war/article-778165, ac- cessed: 2024-07-19

2023
[15]

Pérez, J.M., Giudici, J.C., Luque, F.: pysentimiento: A python toolkit for sentiment analysis and socialnlp tasks (2021)

2021
[16]

Semantic Web10(5), 925–945 (2019)

Zhang, Z., Luo, L.: Hate speech detection: A solved problem? the challenging case of long tail on twitter. Semantic Web10(5), 925–945 (2019)

2019