Marian-Andrei Rizoiu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Marian-Andrei Rizoiu has a total of 117 co-authors. Here are the top co-authors by number of shared publications:
- Lexing Xie
The Australian National University
8 shared publications - Emily Booth
University of Technology Sydney
7 shared publications - Julien Velcin
Université de Lyon, Lyon 2, ERIC UR 3083
6 shared publications Julien Velcin
ERIC Laboratory, University Lumière Lyon, Avenue Pierre Mendés-France, Bron Cedex, France
5 shared publications- Pio Calderon
University of Technology Sydney, Ultimo, NSW, Australia
5 shared publications - Francesco Bailo
University of Sydney
4 shared publications
Latest Publications
- 17 October 2026
- Published by Springer Nature
- p. 88-105
Brexit Means Brexit: Selection Bias, Echo Chambers, and Entrenched Opinion on Reddit
- 17 October 2026
- Published by Springer Nature
- p. 88-105
Brexit Means Brexit: Selection Bias, Echo Chambers, and Entrenched Opinion on Reddit
- Open Access
- 24 September 2026
- Published by Springer Nature in Journal of Computational Social Science
- Vol. 9 (4) , 62
Popping misinformation bubbles: persona-informed style-based messaging in Facebook ad experiments
Misinformation is often viewed as an information problem, attributed to people’s poor understanding of consensus-based facts. While this is a contributing factor, we propose that misinformation and conspiracy are also communicated differently to their audiences. Can we better reach these audiences by matching the styles used in online communities? We present the findings of a multiple Facebook ad study in which we tailored consensus-based messages to communication styles associated with three persona-informed Facebook audiences. Our results demonstrate that styled messages led to higher engagement among targeted Facebook audiences than the neutral-toned variants. We introduce our concept of the misinformation ecosystem, based on our Australian dataset, as a way of understanding online communities and their communication styles, and develop three personas whose demographic and interest profiles informed the Facebook ad audiences used in our experiment. Our findings suggest that styled messaging about consensus-based information can better engage audiences associated with vulnerable communities otherwise resistant to external influence, providing considerable implications about possible future interventions to counter misinformation.
- 24 September 2026
- Published by Springer Nature in Journal of Computational Social Science
- Vol. 9 (4) , 62
Open AccessPopping misinformation bubbles: persona-informed style-based messaging in Facebook ad experiments
Misinformation is often viewed as an information problem, attributed to people’s poor understanding of consensus-based facts. While this is a contributing factor, we propose that misinformation and conspiracy are also communicated differently to their audiences. Can we better reach these audiences by matching the styles used in online communities? We present the findings of a multiple Facebook ad study in which we tailored consensus-based messages to communication styles associated with three persona-informed Facebook audiences. Our results demonstrate that styled messages led to higher engagement among targeted Facebook audiences than the neutral-toned variants. We introduce our concept of the misinformation ecosystem, based on our Australian dataset, as a way of understanding online communities and their communication styles, and develop three personas whose demographic and interest profiles informed the Facebook ad audiences used in our experiment. Our findings suggest that styled messaging about consensus-based information can better engage audiences associated with vulnerable communities otherwise resistant to external influence, providing considerable implications about possible future interventions to counter misinformation.
- 18 June 2026
- Published by Edward Elgar Publishing
- p. 79-104
Designing and running collaborative research projects with Wikibase
This chapter focuses on practical aspects of using Wikibase for collaborative research projects, covering data entry, navigation, querying, and computational access to data. The chapter demonstrates how Wikibase's flexible architecture allows teams to begin entering data without predefining comprehensive structures, enabling schemas and ontologies to develop organically over time. Using detailed examples of political communication research involving Members of Parliament, the chapter illustrates how mixed-methods projects can accommodate diverse data types—from parliamentary transcripts and social media posts to interview recordings—through both convergent and sequential research designs. The chapter explores manual navigation techniques for discovering patterns in linked data, introduces SPARQL querying for systematic data retrieval and analysis, and examines programmatic access methods for automation and large-scale processing. A real-world case study demonstrates how human-in-the-loop machine learning approaches can scale manual coding processes whilst maintaining research rigour.
- 18 June 2026
- Published by Edward Elgar Publishing
- p. 79-104
Designing and running collaborative research projects with Wikibase
This chapter focuses on practical aspects of using Wikibase for collaborative research projects, covering data entry, navigation, querying, and computational access to data. The chapter demonstrates how Wikibase's flexible architecture allows teams to begin entering data without predefining comprehensive structures, enabling schemas and ontologies to develop organically over time. Using detailed examples of political communication research involving Members of Parliament, the chapter illustrates how mixed-methods projects can accommodate diverse data types—from parliamentary transcripts and social media posts to interview recordings—through both convergent and sequential research designs. The chapter explores manual navigation techniques for discovering patterns in linked data, introduces SPARQL querying for systematic data retrieval and analysis, and examines programmatic access methods for automation and large-scale processing. A real-world case study demonstrates how human-in-the-loop machine learning approaches can scale manual coding processes whilst maintaining research rigour.
- 18 June 2026
- Published by Edward Elgar Publishing
- p. 105-117
Collecting, mapping, and modelling social media data with Wikibase
- 18 June 2026
- Published by Edward Elgar Publishing
- p. 105-117
Collecting, mapping, and modelling social media data with Wikibase
- Open Access
- 5 February 2026
- Published by SAGE Publications in New Media & Society
Mapping the moral domain in a modern town hall: Moral foundations’ effects on collective action conditioned by ideology and issue
We examine a year of political debate in a modern town hall to understand the following: when does moral talk turn to talk about action? For which ideological groupings do specific moral foundations matter and in the context of which issues? We analysed 210,170 tweets about the Australian television programme Q&A – a panel show with synchronous live and online audiences. We examined the topics, prevalence of different moralities and how these moralities explain talk of collective action for different ideological groups. Across all issues and ideological groups, talk of action was most often associated with talk of authority but ideological differences between left- and right-leaning participants emerged. Among left-leaning people, action was motivated by a range of moral foundations, but among right-leaning people action was primarily motivated by authority. Our findings map the ideological diversity found within modern town halls and confirm the critical role of these settings in contemporary socio-political life.
- 5 February 2026
- Published by SAGE Publications in New Media & Society
Open AccessMapping the moral domain in a modern town hall: Moral foundations’ effects on collective action conditioned by ideology and issue
We examine a year of political debate in a modern town hall to understand the following: when does moral talk turn to talk about action? For which ideological groupings do specific moral foundations matter and in the context of which issues? We analysed 210,170 tweets about the Australian television programme Q&A – a panel show with synchronous live and online audiences. We examined the topics, prevalence of different moralities and how these moralities explain talk of collective action for different ideological groups. Across all issues and ideological groups, talk of action was most often associated with talk of authority but ideological differences between left- and right-leaning participants emerged. Among left-leaning people, action was motivated by a range of moral foundations, but among right-leaning people action was primarily motivated by authority. Our findings map the ideological diversity found within modern town halls and confirm the critical role of these settings in contemporary socio-political life.
- Open Access
Beyond Content: Behavioral Policies Reveal Actors in Information Operations
The detection of online influence operations -- coordinated campaigns by malicious actors to spread narratives -- has traditionally depended on content analysis or network features. These approaches are increasingly brittle as generative models produce convincing text, platforms restrict access to behavioral data, and actors migrate to less-regulated spaces. We introduce a platform-agnostic framework that identifies malicious actors from their behavioral policies by modeling user activity as sequential decision processes. We apply this approach to 12,064 Reddit users, including 99 accounts linked to the Russian Internet Research Agency in Reddit's 2017 transparency report, analyzing over 38 million activity steps from 2015-2018. Activity-based representations, which model how users act rather than what they post, consistently outperform content models in detecting malicious accounts. When distinguishing trolls -- users engaged in coordinated manipulation -- from ordinary users, policy-based classifiers achieve a median macro-$F_1$ of 94.9%, compared to 91.2% for text embeddings. Policy features also enable earlier detection from short traces and degrade more gracefully under evasion strategies or data corruption. These findings show that behavioral dynamics encode stable, discriminative signals of manipulation and point to resilient, cross-platform detection strategies in the era of synthetic content and limited data access.
Open AccessBeyond Content: Behavioral Policies Reveal Actors in Information Operations
The detection of online influence operations -- coordinated campaigns by malicious actors to spread narratives -- has traditionally depended on content analysis or network features. These approaches are increasingly brittle as generative models produce convincing text, platforms restrict access to behavioral data, and actors migrate to less-regulated spaces. We introduce a platform-agnostic framework that identifies malicious actors from their behavioral policies by modeling user activity as sequential decision processes. We apply this approach to 12,064 Reddit users, including 99 accounts linked to the Russian Internet Research Agency in Reddit's 2017 transparency report, analyzing over 38 million activity steps from 2015-2018. Activity-based representations, which model how users act rather than what they post, consistently outperform content models in detecting malicious accounts. When distinguishing trolls -- users engaged in coordinated manipulation -- from ordinary users, policy-based classifiers achieve a median macro-$F_1$ of 94.9%, compared to 91.2% for text embeddings. Policy features also enable earlier detection from short traces and degrade more gracefully under evasion strategies or data corruption. These findings show that behavioral dynamics encode stable, discriminative signals of manipulation and point to resilient, cross-platform detection strategies in the era of synthetic content and limited data access.
- Open Access
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 2874-2884
X-Troll: eXplainable Detection of State-Sponsored Information Operations Agents
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 2874-2884
Open AccessX-Troll: eXplainable Detection of State-Sponsored Information Operations Agents
- Open Access
Learning to Make Friends: Coaching LLM Agents toward Emergent Social Ties
Can large language model (LLM) agents reproduce the complex social dynamics that characterize human online behavior -- shaped by homophily, reciprocity, and social validation -- and what memory and learning mechanisms enable such dynamics to emerge? We present a multi-agent LLM simulation framework in which agents repeatedly interact, evaluate one another, and adapt their behavior through in-context learning accelerated by a coaching signal. To model human social behavior, we design behavioral reward functions that capture core drivers of online engagement, including social interaction, information seeking, self-presentation, coordination, and emotional support. These rewards align agent objectives with empirically observed user motivations, enabling the study of how network structures and group formations emerge from individual decision-making. Our experiments show that coached LLM agents develop stable interaction patterns and form emergent social ties, yielding network structures that mirror properties of real online communities. By combining behavioral rewards with in-context adaptation, our framework establishes a principled testbed for investigating collective dynamics in LLM populations and reveals how artificial agents may approximate or diverge from human-like social behavior.
Open AccessLearning to Make Friends: Coaching LLM Agents toward Emergent Social Ties
Can large language model (LLM) agents reproduce the complex social dynamics that characterize human online behavior -- shaped by homophily, reciprocity, and social validation -- and what memory and learning mechanisms enable such dynamics to emerge? We present a multi-agent LLM simulation framework in which agents repeatedly interact, evaluate one another, and adapt their behavior through in-context learning accelerated by a coaching signal. To model human social behavior, we design behavioral reward functions that capture core drivers of online engagement, including social interaction, information seeking, self-presentation, coordination, and emotional support. These rewards align agent objectives with empirically observed user motivations, enabling the study of how network structures and group formations emerge from individual decision-making. Our experiments show that coached LLM agents develop stable interaction patterns and form emergent social ties, yielding network structures that mirror properties of real online communities. By combining behavioral rewards with in-context adaptation, our framework establishes a principled testbed for investigating collective dynamics in LLM populations and reveals how artificial agents may approximate or diverge from human-like social behavior.
- Open Access
Mario at EXIST 2025: A Simple Gateway to Effective Multilingual Sexism Detection
This paper presents our approach to EXIST 2025 Task 1, addressing text-based sexism detection in English and Spanish tweets through hierarchical Low-Rank Adaptation (LoRA) of Llama 3.1 8B. Our method introduces conditional adapter routing that explicitly models label dependencies across three hierarchically structured subtasks: binary sexism identification, source intention detection, and multilabel sexism categorization. Unlike conventional LoRA applications that target only attention layers, we apply adaptation to all linear transformations, enhancing the model's capacity to capture task-specific patterns. In contrast to complex data processing and ensemble approaches, we show that straightforward parameter-efficient fine-tuning achieves strong performance. We train separate LoRA adapters (rank=16, QLoRA 4-bit) for each subtask using unified multilingual training that leverages Llama 3.1's native bilingual capabilities. The method requires minimal preprocessing and uses standard supervised learning. Our multilingual training strategy eliminates the need for separate language-specific models, achieving 1.7-2.4\% F1 improvements through cross-lingual transfer. With only 1.67\% trainable parameters compared to full fine-tuning, our approach reduces training time by 75\% and model storage by 98\%, while achieving competitive performance across all subtasks (ICM-Hard: 0.6774 for binary classification, 0.4991 for intention detection, 0.6519 for multilabel categorization).
Open AccessMario at EXIST 2025: A Simple Gateway to Effective Multilingual Sexism Detection
This paper presents our approach to EXIST 2025 Task 1, addressing text-based sexism detection in English and Spanish tweets through hierarchical Low-Rank Adaptation (LoRA) of Llama 3.1 8B. Our method introduces conditional adapter routing that explicitly models label dependencies across three hierarchically structured subtasks: binary sexism identification, source intention detection, and multilabel sexism categorization. Unlike conventional LoRA applications that target only attention layers, we apply adaptation to all linear transformations, enhancing the model's capacity to capture task-specific patterns. In contrast to complex data processing and ensemble approaches, we show that straightforward parameter-efficient fine-tuning achieves strong performance. We train separate LoRA adapters (rank=16, QLoRA 4-bit) for each subtask using unified multilingual training that leverages Llama 3.1's native bilingual capabilities. The method requires minimal preprocessing and uses standard supervised learning. Our multilingual training strategy eliminates the need for separate language-specific models, achieving 1.7-2.4\% F1 improvements through cross-lingual transfer. With only 1.67\% trainable parameters compared to full fine-tuning, our approach reduces training time by 75\% and model storage by 98\%, while achieving competitive performance across all subtasks (ICM-Hard: 0.6774 for binary classification, 0.4991 for intention detection, 0.6519 for multilabel categorization).
- Open Access
- 7 June 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the International AAAI Conference on Web and Social Media
- Vol. 19, 1630-1648
Practical Guidelines for Ideology Detection Pipelines and Psychosocial Applications
Online ideology detection is crucial for downstream tasks, like countering ideologically motivated violent extremism and modeling opinion dynamics. However, two significant issues arise in practitioners' deployment. Firstly, gold-standard training data is prohibitively labor-intensive to collect and has limited reusability beyond its collection context (i.e., time, location, and platform). Secondly, to circumvent expense, researchers employ ideological signals (such as hashtags shared). Unfortunately, these signals' annotation requirements and context transferability are largely unknown, and the bias they induce remains unquantified. This study provides guidelines for practitioners requiring real-time detection of left, right, and extreme ideologies in large-scale online settings. We propose a framework for pipeline constructions, describing ideology signals by their associated labor and context transferability. We evaluate many constructions, quantifying the bias associated with signals and describing a pipeline that outperforms state-of-the-art methods (0.95 AUC ROC). We showcase the capabilities of our pipeline on five datasets containing more than 1.12 million users. We set out to investigate whether the findings in the psychosocial literature, developed for the offline environment, apply to the online setting. We evaluate at scale several psychosocial hypotheses that delineate ideologies concerning morality, grievance, nationalism, and dichotomous thinking. We find that right-wing ideologies use more vice-moral language, have more grievance-filled language, exhibit increased black-and-white thinking patterns, and have a greater association with national flags. This research empowers practitioners with guidelines for ideology detection, and case studies for its application, fostering a safer and better understood digital landscape.
- 7 June 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the International AAAI Conference on Web and Social Media
- Vol. 19, 1630-1648
Open AccessPractical Guidelines for Ideology Detection Pipelines and Psychosocial Applications
Online ideology detection is crucial for downstream tasks, like countering ideologically motivated violent extremism and modeling opinion dynamics. However, two significant issues arise in practitioners' deployment. Firstly, gold-standard training data is prohibitively labor-intensive to collect and has limited reusability beyond its collection context (i.e., time, location, and platform). Secondly, to circumvent expense, researchers employ ideological signals (such as hashtags shared). Unfortunately, these signals' annotation requirements and context transferability are largely unknown, and the bias they induce remains unquantified. This study provides guidelines for practitioners requiring real-time detection of left, right, and extreme ideologies in large-scale online settings. We propose a framework for pipeline constructions, describing ideology signals by their associated labor and context transferability. We evaluate many constructions, quantifying the bias associated with signals and describing a pipeline that outperforms state-of-the-art methods (0.95 AUC ROC). We showcase the capabilities of our pipeline on five datasets containing more than 1.12 million users. We set out to investigate whether the findings in the psychosocial literature, developed for the offline environment, apply to the online setting. We evaluate at scale several psychosocial hypotheses that delineate ideologies concerning morality, grievance, nationalism, and dichotomous thinking. We find that right-wing ideologies use more vice-moral language, have more grievance-filled language, exhibit increased black-and-white thinking patterns, and have a greater association with national flags. This research empowers practitioners with guidelines for ideology detection, and case studies for its application, fostering a safer and better understood digital landscape.