Search Results (93)

Sarcasm detection remains a challenging task in natural language processing because sarcastic expressions often convey meanings that contradict their literal wording. Although transformer-based encoders such as RoBERTa capture contextual semantics effectively, sparse linguistic signals common in sarcastic user-generated text, such as exaggerated punctuation, elongated words, capitalization, and sentiment contrast, may not always remain explicitly accessible in the final sentence representation. To address this limitation, we propose HYSARD, a hybrid feature-fusion model that combines RoBERTa-based sentence embeddings with complementary linguistic features, including sentiment polarity, stylistic markers, syntactic patterns, and TF-IDF lexical cues. The resulting feature space is refined through Random Forest-based feature selection to reduce redundancy and improve robustness, while SMOTE mitigates class imbalance during training. We evaluate HYSARD on the SemEval-2022 iSarcasmEval dataset and the balanced Main and Political subsets of SARC 2.0. Results show strong and consistent performance across datasets, with an F1-score of 0.80 on iSarcasmEval, while held-out test-set error analysis further highlights strong class-wise discrimination. The ablation study further confirms that combining contextual embeddings with explicit linguistic cues improves sarcasm detection over reduced feature configurations. These findings show that hybrid feature fusion remains an effective and practical strategy for sarcasm detection in noisy social media text. Full article

This study investigates the communicative power of editorial cartoons in Ghana’s public discourse, focusing on how they inform, critique, and influence sociopolitical narratives. Drawing on a dataset of cartoons by Tilapia—one of the country’s leading cartoonists—published between May 2024 and May 2025, the paper explores how cartoons address themes such as economic hardship, youth addiction, cultural values, environmental degradation, and political hypocrisy. The central question guiding this study is as follows: How do Tilapia’s editorial cartoons visually construct and critique key national issues—such as economic hardship, environmental degradation, youth addiction, and political hypocrisy—in Ghanaian public discourse? Guided by an integrated theoretical framework from semiotics, visual rhetoric, and critical metaphor theory, the analysis reveals how cartoons use humour, caricature, exaggeration, and symbolic imagery to simplify complex realities and foster civic reflection. The study highlights how cartoons serve not only to entertain but also to hold power to account, amplify public concerns, and promote sociopolitical engagement. Through detailed visual analysis of ten selected cartoons, the paper underscores their capacity to critique governance, expose contradictions, and reflect collective sentiment—especially during election cycles. Overall, the research affirms the evolving role of visual satire as a potent medium of resistance, cultural expression, and democratic participation in Ghana. By bridging visual culture and critical discourse, the paper contributes to broader understandings of the role of the media in shaping public perception and fostering informed citizenship. Full article

Digital social platforms have transformed political discourse into complex socio-technical environments characterized by rapid narrative diffusion, emotional amplification, and large-scale audience interaction. Understanding how sentiment and semantic alignment interact within such environments is important for analyzing polarization and patterns of audience response. This study examines narrative–audience interaction in Telegram political ecosystems using a combination of sentiment analysis, semantic similarity measures, and engagement metrics. Transformer-based language models are applied to quantify relationships between source posts and user-generated comments, enabling joint analysis of affective tone and topical alignment. The results reveal a consistent affective–semantic asymmetry: user responses tend to remain semantically aligned with source narratives while shifting toward more negative sentiment. This pattern indicates that disagreement is predominantly expressed through affective reframing rather than through divergence from the original topic. Further analysis shows systematic differences across ideological groups. Pro-government channels exhibit higher reach and more stable discourse alignment, while pro-opposition channels generate stronger engagement and more pronounced negative sentiment shifts. Neutral channels display intermediate characteristics. These findings demonstrate that online political discourse in Telegram is characterized by stable topical anchoring combined with systematic variation in emotional response. Full article

In the context of political and financial market turmoil, effectively forecasting financial market trends is crucial for investment decisions. Large language models (LLMs) have been applied in extant research to predict market trends, analyze investor sentiments and interpret financial news, all aiming to help investment decision making. However, LLMs face limitations due to training data heterogeneity, restricting multidimensional perspectives and hindering comparative analysis for optimization. This study proposes a “Dual-Agent LLM Debate Mechanism” framework using a Proponent (LLM1: Gemini Pro 3) and an Opponent (LLM2: ChatGPT 5.2) to address single-LLM forecasting gaps: The Proponent generates a baseline forecast (F1) from an Integrated Context, while the Opponent validates and resolves conflicts with the Proponent via up to three rounds of cross-debate to produce a consensus forecast (F2). A controlled experiment was conducted to analyze 75 financial market indicators (FMIs) across five asset categories, revealing that F2 outperforms F1 in accuracy and directional stability, particularly in highly volatile assets like Cryptocurrencies and 10-Year Government Bonds. Paired-sample t-tests confirmed statistical significance, validating the mechanism’s effectiveness. Our study results demonstrate how cross-debate between LLMs enhances forecasting accuracy through structured optimization. Full article

This study examined how Twitter/X users engaged in the political discourse on the Sewol ferry accident in South Korea. We used a triangulation method by combining a social networks approach with quantitative content analysis. A comparison of the number of links across politically homogeneous clusters with the number of links across heterogeneous clusters revealed that selective exposure occurred on the Twitter topic network. Findings also showed the greater role of independent journalists armed with social media in disseminating information online. Our content analysis indicated that the tragic accident divided the public into two sides over the issue and that the public sentiment was dependent on the political orientations of the clusters within the network. The implications of these findings were discussed for scholars who aim to address the problems rooted in a polarized society. Full article

This article examines Romanian internet memes as cultural micro-narratives that encode social critique, identity negotiation, and emotional response through compressed, multimodal storytelling. Using a mixed-method approach, the study integrates qualitative narrative analysis with quantitative sentiment data drawn from the RoMEMESv2 corpus, comprising 983 Romanian-language memes. The analysis identifies recurrent narrative roles and plot structures adapted from Propp’s morphology and applied to digital contexts, revealing archetypal roles, such as the slacker hero, the bureaucratic villain, the domestic guardian, and the trickster. From a quantitative point of view, the corpus exhibits a dominant negative sentiment, particularly within political memes, which combine systemic critique with affective ambivalence. These findings distinguish Romanian memes from datasets in other languages, suggesting that negativity functions not as deviance, but as a culturally specific narrative and emotional resource. Multimodal analysis demonstrates how visual and textual elements operate through anchorage, intertextuality, and symbolic compression, so as to construct narrative messages within single frames. The study argues that Romanian memes function as digital folklore: they narrate social frustration and institutional distrust through irony, repetition, and archetypal condensation, offering insights into the emotional and narrative logic of post-communist digital culture. Full article

The article examines how framing and actor identity structure attention in short-video politics using a country-level corpus from Ecuador. It assembles 4612 public TikTok videos from official accounts and politically salient hashtags, extracts multimodal text via automatic speech recognition and on-screen OCR, and constructs two continuous indices: a quality index (programmatic, efficacy-oriented content) and a populism index (antagonistic, people-versus-elite cues). Engagement is modeled as a fractional response (binomial GLM with logit link), with robustness checks using OLS on logit(ER) and Poisson counts with an offset for log(plays + 1). Models include affect (positive sentiment and anger), hour/day controls, and actor fixed effects (leader, creator, institution, party, and media). The indices display construct validity: quality aligns with positive/joyful tone and populism with anger. Net of controls, populism is positively and consistently associated with engagement across estimators; quality is small and often null or negative. Effects are heterogeneous: leaders gain under both frames, creators primarily under populism, and media modestly under populism, while institutions face penalties under both, and parties show limited returns. Monthly series reveal event-linked intensification of populism, and hashtag networks are modular, mapping onto institutional, partisan, and creator ecosystems. A design analysis identifies a non-populist pathway—benefit-first micro-explanations, concise captions, targeted hashtags, and joyful/efficacy affect—that raises engagement without antagonism. The study contributes a reproducible, open-source pipeline for survey-free, multimodal framing measurement and clarifies how persona × frame interactions and meso-level discursive structure jointly organize attention in short-video politics. Full article

Detecting fake news is essential in natural language processing to verify news authenticity and prevent misinformation-driven social, political, and economic disruptions targeting specific groups. A major challenge in multimodal fake news detection is effectively integrating textual and visual modalities, as semantic gaps and contextual variations between images and text complicate alignment, interpretation, and the detection of subtle or blatant inconsistencies. To enhance accuracy in fake news detection, this article introduces an ensemble-based framework that integrates textual and visual data using ViLBERT’s two-stream architecture, incorporates VADER sentiment analysis to detect emotional language, and uses Image–Text Contextual Similarity to identify mismatches between visual and textual elements. These features are processed through the Bi-GRU classifier, Transformer-XL, DistilBERT, and XLNet, combined via a stacked ensemble method with soft voting, culminating in a T5 metaclassifier that predicts the outcome for robustness. Results on the Fakeddit and Weibo benchmarking datasets show that our method outperforms state-of-the-art models, achieving up to 96% and 94% accuracy in fake news detection, respectively. This study highlights the necessity for advanced multimodal fake news detection systems to address the increasing complexity of misinformation and offers a promising solution. Full article

In this article, the authors address the Serbian media narrative about the EU’s communication on lithium mining in Serbia. In an effort to answer the question of how this narrative can influence the positioning of the EU on Serbia as a candidate country for EU membership, the authors have made a research based on a quantitative–qualitative analysis of media coverage, drawing on a sample of 192 articles (N = 192) published by four Serbian online news portals (RTS, N1, B92, and Blic). The analysis leads to two main conclusions: (1) It indicates an inversion in the general approach to foreign policy orientation across the analyzed media platforms. The customary discourses on Serbia’s foreign policy trajectory temporarily diverged from established patterns—specifically, the fervently pro-Western orientation characteristic of anti-government platforms and the ostensibly West-sceptical orientation typical of pro-government media. This reinforces the argument that the primary structuring line of media discourse in Serbia lies in the division between pro-regime and anti-regime orientations. (2) Media repositioning has exerted a pronounced negative effect on pro-European segments of the Serbian public, reactivating the thesis of “stabilocracy”, conceptualized as the dynamic relationship between authoritarian regimes in the Balkans and their external supporters. According to the authors, the EU’s inability to anticipate the drastic negative shift in public sentiment toward it—particularly among those segments of Serbian society that had been most supportive—or, alternatively, its decision to continue pursuing its own economic interests despite such awareness, underscores the profound flaws in the political communication it employed in this case. Full article

This paper presents a hybrid architecture for automatic sentiment analysis of Kazakh-language political discourse. The Kazakh language is characterized by an agglutinative structure, a complex word-formation system, and the limited availability of digital resources, which significantly complicates the application of standard neural network approaches. To account for these characteristics, a multi-level system was developed that combines morphological and syntactic analysis rules, ontological relationships between political concepts, and multilingual representations of the XLM-R model, used in zero-shot mode. A corpus of 12,000 sentences was annotated for sentiment polarity and used for training and evaluation, while Universal Dependencies annotation was applied for morpho-syntactic analysis. Rule-based components compensate for errors related to affixation variability, modality, and directive constructions. An ontology comprising over 300 domain concepts ensures the correct interpretation of set expressions, terms, and political actors. Experimental results show that the proposed hybrid architecture outperforms both neural network baseline models and purely rule-based solutions, achieving Macro-F1 = 0.81. Ablation revealed that the contribution of modules is unevenly distributed: the ontology provides +0.04 to Macro-F1, the UD syntax +0.08, and the rule-based module +0.11. The developed system forms an interpretable and robust assessment of tonality, emotions, and discursive strategies in political discourse, and also creates a basis for further expansion of the corpus, additional training of models, and the application of hybrid methods to other tasks of analyzing low-resource languages. Full article

This research addresses a critical gap in the Public–Private Partnership (PPP) research field by analysing risk allocation in an emergent African context: Morocco. Based on semi-structured interviews with six selected practitioners, along with lexicometric and sentiment analysis, this study identifies the major risks and the determinants influencing their allocation. Findings show a risk profile dominated by commercial, political, and industrial uncertainties. In addition, the research uncovers that risk allocation is not simply a technical task, but a multidimensional negotiation influenced by project characteristics, partner capabilities, macro-environmental imperatives, and transaction dynamics. Moreover, sentiment analysis reveals a vocabulary mainly reflecting the emotions of fear, anticipation, and trust, which points to the affective side of the contract. This study provides a qualitative framework that is sensitive to the context and that challenges standard economic models; it gives clear directions to policymakers handling complicated PPP arrangements in emerging markets. Full article

Following the Itaewon Halloween Crowd Crush of 29 October 2022, this study examines how public sentiment evolved on Naver, South Korea’s most influential digital platform. While prior research has focused on mainstream media and global social networks, little is known about localized discourse on Naver. To address this gap, we analyzed 2107 user-generated posts collected via Python-based web scraping across three time periods: the immediate aftermath, first anniversary, and passage of the Itaewon Special Law. Semantic network analysis, sentiment classification, and logistic regression were applied to uncover patterns in discourse and emotional tone. Results reveal a shift from grief and outrage in 2022 to demands for political accountability, safety reform, and memorialization by 2024. High-frequency keywords reflected media and government narratives, while low-frequency terms exposed grassroots voices and emotional nuance. Regression analysis confirmed statistically significant associations between sentiment, title length, and year. These findings suggest that digital platforms not only mirror public sentiment but also shape the emotional and political framing of national tragedies. By tracing sentiment over time, this study contributes to understanding how echo chambers, narrative framing, and temporal context interact in shaping collective responses to crisis. Full article

This study investigates the complex dynamics of public discourse surrounding Confucius Institutes (CIs) across the hybrid social networks of mainstream news and social platforms from 2010 to 2023. Employing a longitudinal, multi-platform design, we analyzed news articles and tweets using a computational framework combining topic modeling and sentiment analysis. Our results reveal a shared cross-platform narrative evolution from a “culture-first” to a “politics-central” orientation. However, the trajectory differed significantly: mainstream media underwent a gradual, policy-oriented shift, while social media exhibited an abrupt, nonlinear transition. Crucially, we identify an asymmetric interdependence: Twitter sentiment reliably Granger-causes mainstream media sentiment, establishing its role as a leading indicator, and systematic asymmetries in thematic framing reflect the divergent logics of each platform. The study demonstrates that public discourse on contested, state-linked institutions operates as a complex adaptive system, where bottom-up affective reactions and top-down editorial processes continuously interact in a dynamic equilibrium, ultimately co-constructing a fragmented public understanding. Full article

This study innovatively moves beyond traditional mono-method research by employing an integrated approach that synergizes corpus linguistics and text mining. Through sentiment, thematic, and collocational analyses, it critically examines the representation of China’s image in European media coverage of the China–EU electric vehicle dispute. Initially, sentiment analysis of news reports concerning EU tariffs on Chinese electric vehicles was conducted. Subsequently, four key themes emerged from analyzing a corpus consisting of 202 news articles: “market reaction,”; “trade war,” “China’s response,” and “dialogue and negotiation.” Finally, collocation analysis of the keywords “China” and “Beijing” reveals four main images of China in European media: China is framed as the unfair-subsidy provider, threatener, negotiator, and defender. The key conclusion is that European media coverage is characterized by discursive ambivalence, simultaneously portraying China as both a threat and a partner. These findings are significant as they illuminate how media discourse serves as a key arena where the economic and political complexities of the China–EU trade conflict are negotiated, legitimized, and managed. Full article

The decontextualization (the portrayal of wildlife removed from their natural ecological context through social media), can distort the public perception of these animals and harm conservation efforts. This paper presents an exploratory case study based on two highly visible Facebook videos. To explore this, we analyzed Facebook comments (n = 720) and emoji-based reactions (n = 23,024) regarding Barbary macaques (Macaca sylvanus) in two contexts: entertainment (macaque dressed in sports attire during political protests) and natural habitat (macaque being fed soda by tourists in its forest environment). This is the first study to examine how social media context influences public perception of Barbary macaque conservation status and welfare through analysis of viewer engagement on viral videos. The results indicated that videos depicting macaques in their natural habitat elicited significantly more positive conservation sentiments (68.4% of comments) compared to entertainment contexts (6.04% of comments). Conversely, the entertainment video generated predominantly negative conservation sentiments (54.95% of comments), with viewers expressing amusement rather than concern for species protection. Videos showing macaques in natural settings, particularly when depicting problematic feeding behaviors, prompted more critical engagement and awareness of conservation issues. This pattern suggests that anthropomorphized contexts may obscure recognition of species threats and normalize inappropriate human–wildlife interactions. Given the small dataset, these findings should be interpreted cautiously and as illustrative rather than generalizable. These findings lend preliminary support to the animal decontextualization hypothesis and underscore the importance of context in shaping public perceptions of wildlife and conservation priorities. Full article