Against Political Polarization: A Unified Framework for Tracing Evolving Political Ideologies on Social Media

Yijie Xu; Chao Wang; Hui Xiong

cs.SIAug 18, 2026

Against Political Polarization: A Unified Framework for Tracing Evolving Political Ideologies on Social Media

Authors: Yijie Xu, Chao Wang, Hui Xiong

Organizations: Thrust of Artificial Intelligence, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China · School of Artificial Intelligence and Data Science, University of Science and Technology of China, Hefei, China · Department of Computer Science and Engineering, The Hong Kong University of Science and Technology, Hong Kong, Hong Kong SAR, China

Abstract

The rapid growth of social media has greatly influenced political discourse, highlighting the need to understand individual political ideologies and their temporal dynamics. This task faces challenges such as data scarcity, abundant non-political content, costly and bias-prone manual annotation, and difficulty in modeling future ideological inclinations. To address these issues, we propose TSN4PI, a unified framework for tracking the evolution of political ideologies on social media. It includes two core modules. The PIDN uses large language models with style transfer and unsupervised domain adaptation to enable robust ideology detection and filter irrelevant content from noisy, cross-domain data. The PIPN employs temporal graph neural networks to predict future ideological shifts, enabling comprehensive analysis of ideology presence, intensity, and evolution. We release two large-scale datasets for noncommercial research use to facilitate further work. Extensive case studies on multiple platforms (X and Truth Social) validate the effectiveness of TSN4PI and provide empirical insights into political polarization and the evolution of online ideologies. Our findings offer a nuanced perspective, advancing both methodological development and empirical understanding in this field.

Figures & tables

Figure 1. Illustration of political ideology dynamics. (a)–(b) Original vector mockups of contrasting X-style and Truth Social-style posts; identities, text, avatars, and engagement counts are illustrative. (c) Temporal graph snapshots. (d) Original five-category scales of AllSides-derived outlet scores. Four original vector panels illustrate political ideology dynamics: two asset-free mockups of contrasting social posts, directed-network snapshots at time steps 1 and 2, and custom scales placing ABC News at -2.4 (Lean Left) and Fox News at 4.0 (Right). Scale labels L, LL, C, LR, and R denote Left, Lean Left, Center, Lean Right, and Right. No platform logos or profile photographs are used.

Figure 2. Distribution of articles (left) and outlets (right) by political leaning, based on data from AllSides. The bar graph shows the top 20 sources by share of total articles, colored by ideology. Two pie charts show news-article and media-outlet shares across five political leanings, with Center the largest category in both. A bar chart below ranks the top 20 outlets by article share, with The Hill highest.

Rank	Left	Lean Left	Center	Lean Right	Right
1	trump (96,727)	people (130,164)	state (122,494)	president (54,261)	trump (60,993)
2	people (62,773)	president (96,049)	trump (121,089)	biden (45,403)	biden (50,717)
3	election (58,163)	state (94,501)	president (115,675)	house (42,055)	house (50,398)
4	republican (47,839)	new (92,173)	people (109,190)	new (41,471)	president (47,401)
5	even (47,647)	trump (88,297)	new (105,062)	state (40,075)	news (47,398)
6	president (46,673)	covid (86,424)	house (95,068)	congress (34,186)	people (40,715)

Table 1. Top 10 most frequent content words per political leaning (full text, stopwords removed).

Table 2. X Dataset Overview.

Dataset Name	Platform	Count	Users	Duration	Data Level	Theme
Dreaddit ( Turcan and McKeown, 2019 )	Reddit	$\sim$ 190k	N/A	2017–2018	Post	Social media stress
Reddit Corpus ( Henderson et al., 2019 )	Reddit	727M	N/A	2015–2018	Dialogue	Conversational AI
Weibo-COV ( Hu et al., 2020 )	Sina Weibo	>40M	20M	2019–2020	Post	COVID-19 discourse
News Feed Rank. ( Boukhalfa et al., 2022 )	Twitter	26k	46	2021–2022	News	Personalized ranking
MetaHate ( Piot et al., 2024 )	Mixed	1.2M	N/A	N/A	Post	Hate speech detect.
IsamasRed ( Chen et al., 2024b )	Reddit	8M	N/A	2023	Dialogue	Israel-Hamas discourse

Table 3. Summary of Recent Social Media Datasets

Panel of Figure 3 — Figure 3. Temporal growth and topical focus of the X (Twitter) dataset. Two panels summarize the X dataset. The left line chart shows monthly tweet counts rising from near zero in 2007 to roughly two million by late 2022, with intermediate spikes. The right horizontal bar chart ranks 30 hashtags, led by p2, trump, news, smartnews, and socialmedia.

Field	# of Records
Users	454,458
Truths	823,927
Quotes	10,508
Replies	506,276

Table 4. Truth Social Dataset Overview

Figure 4. The overview of our proposed framework: TSN4PI . The upper part of the figure represents the PIDN module, while the lower part represents the PIPN module. A two-stage pipeline. The upper PIDN transforms news into social-media style, combines it with daily-life and unlabeled social-media posts, encodes them with a BERT-based language model, and optimizes classification, regression, and domain-alignment losses. The lower PIPN processes temporal social graphs with a temporal graph neural network for link prediction and node classification and regression.

Figure 5. Embedding across PLMs. Four UMAP scatter plots compare AllSides news, social-media posts, and style-transferred news under BERT, XLM-R, Longformer, and TWHIN-BERT embeddings. Across models, the style-transferred news shifts toward the social-media cluster.

	Classification (F1 $\uparrow$ )			Political Class Detail $\uparrow$		Regression Error $\downarrow$
Model / Setting	Non-political	Political	Macro Avg.	Precision	Recall	MAE	RMSE
Cross-Model Scaling (0-shot)
Qwen2.5-0.5B-Instruct	0.510 $\,\pm\,$ 0.022	0.727 $\,\pm\,$ 0.007	0.618 $\,\pm\,$ 0.014	0.601 $\,\pm\,$ 0.007	0.918 $\,\pm\,$ 0.006	0.413 $\,\pm\,$ 0.007	0.478 $\,\pm\,$ 0.005
Qwen2.5-3B-Instruct	0.946 $\,\pm\,$ 0.003	0.943 $\,\pm\,$ 0.003	0.945 $\,\pm\,$ 0.003	0.945 $\,\pm\,$ 0.008	0.942 $\,\pm\,$ 0.006	0.283 $\,\pm\,$ 0.005	0.333 $\,\pm\,$ 0.005
Qwen2.5-7B-Instruct	0.927 $\,\pm\,$ 0.002	0.905 $\,\pm\,$ 0.003	0.916 $\,\pm\,$ 0.002	0.992 $\,\pm\,$ 0.003	0.833 $\,\pm\,$ 0.004	0.238 $\,\pm\,$ 0.004	0.288 $\,\pm\,$ 0.005
Qwen2.5-14B-Instruct	0.918 $\,\pm\,$ 0.003	0.900 $\,\pm\,$ 0.004	0.909 $\,\pm\,$ 0.003	0.986 $\,\pm\,$ 0.001	0.827 $\,\pm\,$ 0.006	0.227 $\,\pm\,$ 0.004	0.285 $\,\pm\,$ 0.003

Table 5. Unified performance analysis of the Qwen2.5 model family. Top: cross-model scaling analysis in a zero-shot setting. Bottom: few-shot analysis for the Qwen2.5-7B-Instruct model. Within each block, bold marks the best and underlining the second best per metric.

Figure 6. Inference latency comparison. Left: 0-shot latency of PIDN ( $\approx$ 0.3B) vs. LLMs of varying sizes. Right: Latency of PIDN vs. Qwen2.5-7B with increasing ICL shots (0–10). Latencies are reported in milliseconds (ms). The left bar chart shows PIDN latency at 4.39 milliseconds versus Qwen2.5 models ranging from 92.90 milliseconds for 0.5B to 1145.71 milliseconds for 72B. The right plot shows Qwen2.5-7B latency remaining around 341 to 342 milliseconds across zero to ten shots, while PIDN remains at 4.39 milliseconds.

Table 6. Comparison of PLM backbones and hyperparameter tuning results. Best values are in bold .

Configuration	Acc. (Kind) (%) $\uparrow$	MSE (Score) $\downarrow$
Backbone Only	97.79 $\,\pm\,$ 0.15	0.0670 $\,\pm\,$ 0.0012
Backbone + UDA ( KL )	98.15 $\,\pm\,$ 0.11	0.0350 $\,\pm\,$ 0.0009
Backbone + UDA ( MMD )	98.35 $\,\pm\,$ 0.10	0.0320 $\,\pm\,$ 0.0010
Backbone + TST	99.12 $\,\pm\,$ 0.08	0.0060 $\,\pm\,$ 0.0004
Backbone + TST + UDA ( KL ) ( PIDN - KL )	99.25 $\,\pm\,$ 0.07	0.0048 $\,\pm\,$ 0.0003
Backbone + TST + UDA ( MMD ) ( PIDN - MMD )	99.52 $\,\pm\,$ 0.05	0.0057 $\,\pm\,$ 0.0004

Table 7. Ablation study of TST and UDA methods. All experiments use the twhin-bert-base backbone and the selected hyperparameter setting (

\alpha=0.4,\beta=0.3

). Best performance is in bold .

Figure 7. Evaluation of Text Style Transfer. Top: Histograms of similarity scores between original news and tweet-style outputs, computed using four sentence embedding models. Bottom: Histograms of Style Appropriateness and Semantic Preservation scores (normalized to 0–1) assigned by LLM judges. Eight histograms evaluate text style transfer. The top row shows news-to-tweet similarity distributions from four embedding models, concentrated mostly between 0.6 and 0.9. The bottom row shows style and semantic scores from DeepSeek and OpenAI judges, with most probability mass at higher scores.

Dataset	Model	Temporal	Link Prediction		Ideology Prediction
Dataset	Model	Temporal	AP(%) $\uparrow$	AUC(%) $\uparrow$	Acc.(%) $\uparrow$	MSE $\downarrow$
Truth Social	GCN ( Kipf and Welling, 2017 )	✗	N/A	N/A	58.35 $\,\pm\,$ 0.85	0.045 $\,\pm\,$ 0.002
	GAT ( Veličković et al., 2018 )	✗	N/A	N/A	61.81 $\,\pm\,$ 0.72	0.043 $\,\pm\,$ 0.002
	TGN ( Rossi et al., 2020 )	✓	99.25 $\,\pm\,$ 0.08	99.14 $\,\pm\,$ 0.09	72.42 $\,\pm\,$ 0.35	0.036 $\,\pm\,$ 0.001
	JODIE ( Kumar et al., 2019 )	✓	99.73 $\,\pm\,$ 0.04	99.69 $\,\pm\,$ 0.05	66.28 $\,\pm\,$ 0.41	0.039 $\,\pm\,$ 0.001
	APAN ( Wang et al., 2021c )	✓	97.56 $\,\pm\,$ 0.15	97.24 $\,\pm\,$ 0.18	65.14 $\,\pm\,$ 0.45	0.036 $\,\pm\,$ 0.001

Table 8. Comparative results of static (✗) and temporal (✓) models. The evaluation covers Link Prediction (AP, AUC) and Ideology Prediction (Accuracy, MSE). Best performance for each metric is in bold ; second best is underlined . Static models are not applicable (N/A) for the link prediction task.

Panel of Figure 8 — Figure 8. Ideological structure and temporal dynamics across platforms. (a) shows how users on the two platforms are distributed along the left–right scale, which reflects cross-platform differences in baseline audience composition. (b) tracks how highly unstable users move on the same scale over time. Panel (a) compares ideology-score histograms: Twitter is concentrated left of zero near -0.25, whereas Truth Social is concentrated on the positive, right-leaning side. Panel (b) plots yearly mean scores for the 20 most volatile Twitter users, colored by initial ideology, with a dashed line at the 0.5 midpoint.

User	Year	Representative tweet (truncated)	Score
1	2009	Is there such a thing as a “Lie Factory” ? If so, its obvious right-wing Republicans & their talking heads hold a great deal of stock in it.	0.073
		3 days left. Vote 4 my rescue group 2 save lives ! Retweet!	0.369
	2022	As you know, currently the price per barrel of oil is $120 . In this chart, note how low the price has to be for corps to profit by drilling. It’s the greed .	0.515
		Next time a [Republican] points out that more have died from Covid under [President A] than [President B], remind them that correlation doesn’t equal causation .	0.521
		This is where I want my taxes to go, investing in America and future generations . A population without access to higher education or drowning in debt cannot compete.	0.319
2	2017	There’s been some great addresses about #ScriptureStudy . What will you begin to study tomorrow? #[ReligiousConference]	0.321

Table 9. Case study of three Twitter users illustrating non-polarizing trajectories within the U.S. political context. For each user, representative tweets are shown from two distinct periods. Key phrases are bolded .

Year	Left-leaning Topics	Center-leaning Topics	Right-leaning Topics
2009	[hcr, health, reform] [copenhagen, cop15]	[stimulus, job, economy] [hcr, health, reform] [iran, iranelection]	[spending, stimulus, govt] [hcr, gop, tcot] [iran, iranelection]
2010	[health, reform, hcr] [haiti, earthquake]	[oil, bp, oilspill] [haiti] [election, mn2010, tea, teaparty]	[hcr, obamacare, repeal] [tea, party, teaparty] [flgov, scott, spending]
2011	[egypt, libya] [medicare, cuts, deficit] [wiunion]	[debt, debtceiling, boehner] [egypt, mubarak, libya] [bin laden, pakistan]	[libya, nato, war] [wiunion, gov, walker] [debt, obama, gop]
2012	[waronwomen, women] [romney, mitt, returns] [health, medicare]	[obama, romney, debate] [syria] [gun, guncontrol, newtown, nra]	[libya, benghazi, attack] [obama, romney, debate] [gun, shooting, aurora]
2013	[standwithwendy, txlege, abortion] [gun, guns, guncontrol] [blacklivesmatter]	[snowden, leak, nsa] [boston, marathon, bombing] [shutdown, government]	[irs, lerner, targeting] [benghazi, cia] [syria, assad, military]
2014	[ferguson, police, racism] [acaworks, aca] [climate, change]	[ferguson, michael brown] [ukraine, russia, putin] [ebola] [mh370]	[isis, syria, iraq] [benghazi] [bridgegate, christie] [irs, lerner, emails]

Table 10. Representative Topics by Political Leaning and Year (2009–2022).

Year	Left-leaning Topics	Center-leaning Topics	Right-leaning Topics
2022	[abortion, roe, wade, overturned] [uvalde, shooting, gun, control] [ukraine, standwithukraine, kyiv] [jan6, committee, hearings, insurrection] [maralago, raid, fbi]	[ukraine, russia, war, putin] [maralago, raid, fbi, doj, warrant] [roe, wade, overturned, supreme, court] [inflation, recession, economy, prices] [hunter, laptop, bidens, fbi] [2000mules, ballot, harvesting, fraud]	[inflation, gas, prices, bidenflation] [fbi, doj, raid, corruption, witchhunt] [border, crisis, migrants, illegal] [hunter, laptop, biden, crimefamily] [stoptheshots, diedsuddenly, pureblood] [abortion, prolife, unborn, baby]

Table 11. Representative Topics on Truth Social by Political Leaning (2022).

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Characteristic	# of News Articles	# of Media Outlets
Total Articles/Outlets	233,570	466
Political Ideology
Center	76,023	158
Lean Left	53,532	106
Right	45,506	75
Lean Right	32,555	62

Appendix

Table 1. Overview of the allsides.com news with annotated political ideology dataset.

Figure 1. Word Clouds Highlighting Common Terms in News Reports Titles by Political Leanings. The visualization spans from ‘Left’ to ‘Right’, with ‘Center’ in the middle, reflecting the language usage patterns in political news reporting. Five side-by-side word clouds compare title terms for Left, Lean Left, Center, Lean Right, and Right outlets. Prominent phrases include Capitol Riot and Taylor Greene on the left, COVID vaccine and Supreme Court in the center, and criminal referrals and vaccine mandate on the right.

Metric	Time Complexity	FLOPs (Approx.)	GPU Bottleneck
MMD	$O(N^{2}D)$	$9N^{2}D$	Compute-Bound
KL Divergence	$O(ND)$	$4ND$	Memory-Bound

Appendix

Table 5. Computational complexity comparison of UDA methods.

	Classification (F1 $\uparrow$ )			Political Class Detail $\uparrow$		Regression Error $\downarrow$
Model / Setting	Non-political	Political	Macro Avg.	Precision	Recall	MAE	RMSE
Few-Shot for Qwen2.5-0.5B-Instruct
0-shot	0.510 $\,\pm\,$ 0.022	0.727 $\,\pm\,$ 0.007	0.618 $\,\pm\,$ 0.014	0.601 $\,\pm\,$ 0.007	0.918 $\,\pm\,$ 0.006	0.413 $\,\pm\,$ 0.007	0.478 $\,\pm\,$ 0.005
2-shot	0.568 $\,\pm\,$ 0.132	0.717 $\,\pm\,$ 0.035	0.643 $\,\pm\,$ 0.082	0.629 $\,\pm\,$ 0.094	0.855 $\,\pm\,$ 0.061	0.297 $\,\pm\,$ 0.055	0.364 $\,\pm\,$ 0.054
4-shot	0.607 $\,\pm\,$ 0.132	0.751 $\,\pm\,$ 0.051	0.679 $\,\pm\,$ 0.089	0.662 $\,\pm\,$ 0.126	0.901 $\,\pm\,$ 0.088	0.330 $\,\pm\,$ 0.031	0.395 $\,\pm\,$ 0.038
6-shot	0.577 $\,\pm\,$ 0.098	0.750 $\,\pm\,$ 0.032	0.664 $\,\pm\,$ 0.062	0.622 $\,\pm\,$ 0.040	0.951 $\,\pm\,$ 0.052	0.305 $\,\pm\,$ 0.016	0.367 $\,\pm\,$ 0.018

Appendix

Table 6. Unified performance analysis of the Qwen2.5 model family (Part 1: 0.5B, 3B, 7B). All models are evaluated from 0-shot to 200-shot settings.

	Classification (F1 $\uparrow$ )			Political Class Detail $\uparrow$		Regression Error $\downarrow$
Model / Setting	Non-political	Political	Macro Avg.	Precision	Recall	MAE	RMSE
Few-Shot for Qwen2.5-14B-Instruct
0-shot	0.918 $\,\pm\,$ 0.003	0.900 $\,\pm\,$ 0.004	0.909 $\,\pm\,$ 0.003	0.986 $\,\pm\,$ 0.001	0.827 $\,\pm\,$ 0.006	0.227 $\,\pm\,$ 0.004	0.285 $\,\pm\,$ 0.003
2-shot	0.949 $\,\pm\,$ 0.022	0.941 $\,\pm\,$ 0.029	0.945 $\,\pm\,$ 0.026	0.992 $\,\pm\,$ 0.003	0.896 $\,\pm\,$ 0.054	0.219 $\,\pm\,$ 0.007	0.273 $\,\pm\,$ 0.011
4-shot	0.955 $\,\pm\,$ 0.009	0.949 $\,\pm\,$ 0.011	0.952 $\,\pm\,$ 0.010	0.994 $\,\pm\,$ 0.001	0.908 $\,\pm\,$ 0.021	0.216 $\,\pm\,$ 0.002	0.271 $\,\pm\,$ 0.007
6-shot	0.963 $\,\pm\,$ 0.014	0.959 $\,\pm\,$ 0.017	0.961 $\,\pm\,$ 0.016	0.991 $\,\pm\,$ 0.004	0.929 $\,\pm\,$ 0.034	0.235 $\,\pm\,$ 0.025	0.299 $\,\pm\,$ 0.032

Appendix

Table 7. Unified performance analysis of the Qwen2.5 model family (Part 2: 14B, 32B, 72B). All models are evaluated from 0-shot to 200-shot settings.

Social Media Analysis

Social media analysis studies online posts, media, users, and interactions to understand communication, communities, and public behavior.

See all 221 papers

When Words Divide: Diachronic Ideological Polarization in Political Discourse on Social Media
Aug 2, 2026Roy Yitzchak, Noa Lavie, Ella RabinovichSocial Media Analysis Political Discourse Analysis
The Academic College of Tel-Aviv Yaffo, Israel
Quantifying Political Partisanship for Cross-Platform Analyses
Jul 23, 2026Fathima Ameen, Christopher G. HealeySocial Media Analysis Computational Social Science
North Carolina State University, Raleigh, NC, USA
Automated Identification of Competing Narratives in Political Discourse on Social Media
Sep 10, 2026Sergej Wildemann, Erick ElejaldeSocial Media Analysis Political Discourse Analysis
L3S Research Center, Leibniz Universität Hannover, Hannover, Germany

Aug 2, 2026J/K move · Enter open · S save

When Words Divide: Diachronic Ideological Polarization in Political Discourse on Social Media

Roy Yitzchak, Noa Lavie, Ella Rabinovich

The Academic College of Tel-Aviv Yaffo, Israel

Social Media Analysis Political Discourse Analysis Computational Social Science

Open paper AskPDF arXiv

Political polarization has become a defining feature of online discourse, yet its long-term evolution remains poorly understood. We present a longitudinal analysis of ideological polarization in Reddit discussions by measuring semantic differences in the language used by opposing political communities. We construct temporally aligned community-specific word embeddings and quantify ideological polarization as the semantic divergence of political concepts over time. Our analysis shows that ideological polarization has increased substantially during the study period, both at the concept- and topic-level. Unlike prior computational work, which has largely focused on cross-sectional analyses or affective dimensions of polarization at a single point at time, our approach captures the evolution of ideological differences in semantic framing. The proposed framework provides a scalable method for studying the temporal dynamics of ideological polarization in large-scale social media discourse.

Abstract

Social Media Analysis

Explore similar work

When Words Divide: Diachronic Ideological Polarization in Political Discourse on Social Media

Quantifying Political Partisanship for Cross-Platform Analyses

Automated Identification of Competing Narratives in Political Discourse on Social Media