Scholarly Works
Research Output
Recent Publications
Real talk or posed shots? Unpacking GenZs reaction to candid vs curated brand posts on BeReal
Publishing year: 2026
Journal article
EuroMed Journal of Business, 2026, Volume 21, Issue 5
Repository
Pages 96-116
Purpose
This study explores how brands can engage Generation Z on BeReal, a social media platform praised for its authenticity. It examines (1) the psychological drivers of GenZ's platform engagement and (2) how different types of brand posts affect advertising outcomes.
Design/methodology/approach
A multi-method approach was employed in two empirical studies. Study 1 surveyed 208 GenZ users to test the psychological mechanisms (e.g. narcissism and brand selfies) driving self-brand connection. Study 2 conducted a 2 × 2 online experiment to assess the influence of type of post (candid vs curated) and product involvement on ad recognition, ad irritation, brand attitudes and purchase intentions. To contextualize BeReal's cultural footprint, we analyzed the platform's publicly accessible presence on X (formerly Twitter).
Findings
BeReal users span diverse geographies but share a preference for authenticity. Narcissism and brand selfies mediate the relationship between BeReal's use and self-brand connections. Candid brand posts reduce ad recognition and ad irritation, improving brand attitudes and purchase intentions, but only for low-involvement products. These effects do not emerge for high-involvement products, which require more structured and informative messaging.
Research limitation...
Model-Contingent Polarity Bias in Large Language Model Annotation: Implications for Semantic Multimedia Personalization
Publishing year: 2026
Journal article
Computers, 2026, Volume 15, Issue 5
Repository
Large Language Models (LLMs) are increasingly deployed as automated annotators in semantic multimedia systems, yet their reliability varies significantly across architectures. This study extends prior cross-model evaluations by benchmarking ChatGPT-5, Qwen-3, and Gemini-3-flash against human expert annotations using the HRAST hotel review dataset. We adopt a bias-by-design framework to analyze systematic divergences in sentiment, topic, and aspect labeling across real and synthetic data, while investigating the moderating effects of annotation mode. Findings reveal model-contingent polarity bias: ChatGPT-5 exhibits a pronounced neutrality bias, while Qwen-3 and Gemini-3-flash align more closely with human polarization. Agreement is substantial for concrete topics but diverges on abstract evaluative dimensions. Synthetic data consistently inflates reliability metrics while masking ambiguity. These findings highlight that annotation bias is structurally embedded in model design choices and operational conditions. Cross-architectural triangulation and mode-aware deployment strategies are recommended for robust semantic multimedia system development.
Funding the Want, Not the Need: Self-Transformation and Moral Emotions in Desire-Based Crowdfunding
Publishing year: 2026
Conference paper
19th Annual Conference of the EuroMed Academy, 2026
Repository
Can Large Language Models Replace Keyword-Based Mappings for SDG Classification? A Zero-Shot Evaluation Against the Elsevier 2025 Standard
Publishing year: 2026
Journal article
Sustainability Switzerland, 2026, Volume 18, Issue 20
Scopus
Automated classification of texts against the United Nations Sustainable Development Goals (SDGs) is an important capability for academic institutions required to report on their contributions to the 2030 Agenda. Keyword-based mappings such as the Elsevier 2025 SDG Research Mapping are widely adopted in institutional practice, while Large Language Models (LLMs) have recently emerged as a flexible alternative, yet no direct empirical comparison between the two paradigms has been undertaken. We evaluate five state-of-the-art LLMs on multi-label SDG classification over a 5000-text sample from the OSDG Community Dataset, comparing their zero-shot outputs against the Elsevier 2025 mapping. The mapping is adopted as a reference standard rather than as ground truth, so the comparison measures compatibility with institutional practice. The analysis combines a holistic comparison of classification volumes and distributional alignment with a per-SDG analysis, including a scope expansion experiment that replaces exact keyword matching with embedding-based semantic matching. All five LLMs over-classify in aggregate, assigning between 1.15 and 1.81 times as many labels, though not uniformly. Of 85 model-goal combinations, 64 lie significantly above the Elsevier counts and 11 below. Embedding-based matching narrows the gap on nine of the 17 goals but overshoots the models on the remaining eight. As the expansion also removes the Boolean structure of the queries, these results are consisten...
Preface
Publishing year: 2025
Journal article
Proceedings 2025 20th International Workshop on Semantic and Social Media Adaptation and Personalization Smap 2025, 2025
Scopus
Pages xi-xii
Media battles in the cybersphere: analyzing news and social media agendas during the 2015 Greek bailout referendum
Publishing year: 2025
Journal article
Frontiers in Political Science, 2025, Volume 6
Repository
244
Unique publications
29.1%
Open access (71 publications; 68 without information)
27%
With other institutions (66 publications; 85 without information)
H-Index Reports
| Index type | Scopus | Google Scholar | Web of Science |
|---|---|---|---|
| H-index | 20 | 31 | 1 |
| H5-index | 5 | 7 | 1 |
| H10-index | 11 | 14 | 1 |
Publications per year and source
No description yet.
- Faculty
- Faculty of Communication and Media Studies
- Department
- Department of Communication and Marketing
- Title
- Professor