Leon Fröhling
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Leon Fröhling has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Gianluca Demartini
The University of Queensland, Brisbane, Australia
2 shared publications Indira Sen
University of Konstanz, Germany
2 shared publications- Kevin Roitero
University of Udine, udine, Italy
2 shared publications - Pietro Bernardelle
The University of Queensland, Brisbane, Australia
2 shared publications - Riccardo Lunardi
University of Udine, udine, Italy
2 shared publications - Stefano Civelli
The University of Queensland, Brisbane, Australia
2 shared publications
Latest Publications
- Open Access
- 9 September 2026
- Published by Elsevier in Information Processing & Management
- Vol. 64 (2)
Political Ideology Shifts in Large Language Models
- 9 September 2026
- Published by Elsevier in Information Processing & Management
- Vol. 64 (2)
Open AccessPolitical Ideology Shifts in Large Language Models
- Open Access
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-48
Agent-Supported Foresight for AI Systemic Risks: AI Agents for Breadth, Experts for Judgment
AI impact assessments often stress near-term risks because human judgment degrades over longer horizons, exemplifying the Collingridge dilemma: foresight is most needed when knowledge is scarcest. To address long-term systemic risks, we introduce a scalable approach that simulates in-silico agents using the foresight method of the Futures Wheel. We applied it to four AI uses spanning Technology Readiness Levels (TRLs): Chatbot Companion (TRL 9), AI Toy (TRL 7), Griefbot (TRL 5), and Death App (TRL 2). Across 30 agent runs per use, agents produced 86–110 consequences, condensed into 27–47 unique risks. To benchmark the agent outputs against human perspectives, we collected evaluations from 290 domain experts and 7 leaders, and conducted Futures Wheel sessions with 42 experts and 42 laypeople. Agents generated many systemic consequences. Compared with these outputs, experts identified fewer risks, typically less systemic but judged more likely, whereas laypeople surfaced more emotionally salient concerns that were generally less systemic. We propose a hybrid foresight workflow, wherein agents broaden systemic coverage, and humans provide contextual grounding.
- 13 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-48
Open AccessAgent-Supported Foresight for AI Systemic Risks: AI Agents for Breadth, Experts for Judgment
AI impact assessments often stress near-term risks because human judgment degrades over longer horizons, exemplifying the Collingridge dilemma: foresight is most needed when knowledge is scarcest. To address long-term systemic risks, we introduce a scalable approach that simulates in-silico agents using the foresight method of the Futures Wheel. We applied it to four AI uses spanning Technology Readiness Levels (TRLs): Chatbot Companion (TRL 9), AI Toy (TRL 7), Griefbot (TRL 5), and Death App (TRL 2). Across 30 agent runs per use, agents produced 86–110 consequences, condensed into 27–47 unique risks. To benchmark the agent outputs against human perspectives, we collected evaluations from 290 domain experts and 7 leaders, and conducted Futures Wheel sessions with 42 experts and 42 laypeople. Agents generated many systemic consequences. Compared with these outputs, experts identified fewer risks, typically less systemic but judged more likely, whereas laypeople surfaced more emotionally salient concerns that were generally less systemic. We propose a hybrid foresight workflow, wherein agents broaden systemic coverage, and humans provide contextual grounding.
- Open Access
- 8 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 864-867
Mapping and Influencing the Political Ideology of Large Language Models using Synthetic Personas
- 8 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 864-867
Open AccessMapping and Influencing the Political Ideology of Large Language Models using Synthetic Personas
- Open Access
- 13 June 2024
- Published by SAGE Publications in Social Science Computer Review
- Vol. 43 (5) , 1114-1144
The Unseen Targets of Hate: A Systematic Review of Hateful Communication Datasets
Machine learning (ML)-based content moderation tools are essential to keep online spaces free from hateful communication. Yet ML tools can only be as capable as the quality of the data they are trained on allows them. While there is increasing evidence that they underperform in detecting hateful communications directed towards specific identities and may discriminate against them, we know surprisingly little about the provenance of such bias. To fill this gap, we present a systematic review of the datasets for the automated detection of hateful communication introduced over the past decade, and unpack the quality of the datasets in terms of the identities that they embody: those of the targets of hateful communication that the data curators focused on, as well as those unintentionally included in the datasets. We find, overall, a skewed representation of selected target identities and mismatches between the targets that research conceptualizes and ultimately includes in datasets. Yet, by contextualizing these findings in the language and location of origin of the datasets, we highlight a positive trend towards the broadening and diversification of this research space.
- 13 June 2024
- Published by SAGE Publications in Social Science Computer Review
- Vol. 43 (5) , 1114-1144
Open AccessThe Unseen Targets of Hate: A Systematic Review of Hateful Communication Datasets
Machine learning (ML)-based content moderation tools are essential to keep online spaces free from hateful communication. Yet ML tools can only be as capable as the quality of the data they are trained on allows them. While there is increasing evidence that they underperform in detecting hateful communications directed towards specific identities and may discriminate against them, we know surprisingly little about the provenance of such bias. To fill this gap, we present a systematic review of the datasets for the automated detection of hateful communication introduced over the past decade, and unpack the quality of the datasets in terms of the identities that they embody: those of the targets of hateful communication that the data curators focused on, as well as those unintentionally included in the datasets. We find, overall, a skewed representation of selected target identities and mismatches between the targets that research conceptualizes and ultimately includes in datasets. Yet, by contextualizing these findings in the language and location of origin of the datasets, we highlight a positive trend towards the broadening and diversification of this research space.
- Top Cited PapersOpen Access
- 27 April 2024
- Published by SAGE Publications in Social Science Computer Review
- Vol. 43 (5) , 943-979
Assessing Data Quality in the Age of Digital Social Research: A Systematic Review
While survey data has long been the focus of quantitative social science analyses, observational and content data, although long-established, are gaining renewed attention; especially when this type of data is obtained by and for observing digital content and behavior. Today, digital technologies allow social scientists to track “everyday behavior” and to extract opinions from public discussions on online platforms. These new types of digital traces of human behavior, together with computational methods for analyzing them, have opened new avenues for analyzing, understanding, and addressing social science research questions. However, even the most innovative and extensive amounts of data are hollow if they are not of high quality. But what does data quality mean for modern social science data? To investigate this rather abstract question the present study focuses on four objectives. First, we provide researchers with a decision tree to identify appropriate data quality frameworks for a given use case. Second, we determine which data types and quality dimensions are already addressed in the existing frameworks. Third, we identify gaps with respect to different data types and data quality dimensions within the existing frameworks which need to be filled. And fourth, we provide a detailed literature overview for the intrinsic and extrinsic perspectives on data quality. By conducting a systematic literature review based on text mining methods, we identified and reviewed 58 data quality frameworks. In our decision tree, the three categories, namely, data type, the perspective it takes, and its level of granularity, help researchers to find appropriate data quality frameworks. We, furthermore, discovered gaps in the available frameworks with respect to visual and especially linked data and point out in our review that even famous frameworks might miss important aspects. The article ends with a critical discussion of the current state of the literature and potential future research avenues.
- 27 April 2024
- Published by SAGE Publications in Social Science Computer Review
- Vol. 43 (5) , 943-979
Top Cited PapersOpen AccessAssessing Data Quality in the Age of Digital Social Research: A Systematic Review
While survey data has long been the focus of quantitative social science analyses, observational and content data, although long-established, are gaining renewed attention; especially when this type of data is obtained by and for observing digital content and behavior. Today, digital technologies allow social scientists to track “everyday behavior” and to extract opinions from public discussions on online platforms. These new types of digital traces of human behavior, together with computational methods for analyzing them, have opened new avenues for analyzing, understanding, and addressing social science research questions. However, even the most innovative and extensive amounts of data are hollow if they are not of high quality. But what does data quality mean for modern social science data? To investigate this rather abstract question the present study focuses on four objectives. First, we provide researchers with a decision tree to identify appropriate data quality frameworks for a given use case. Second, we determine which data types and quality dimensions are already addressed in the existing frameworks. Third, we identify gaps with respect to different data types and data quality dimensions within the existing frameworks which need to be filled. And fourth, we provide a detailed literature overview for the intrinsic and extrinsic perspectives on data quality. By conducting a systematic literature review based on text mining methods, we identified and reviewed 58 data quality frameworks. In our decision tree, the three categories, namely, data type, the perspective it takes, and its level of granularity, help researchers to find appropriate data quality frameworks. We, furthermore, discovered gaps in the available frameworks with respect to visual and especially linked data and point out in our review that even famous frameworks might miss important aspects. The article ends with a critical discussion of the current state of the literature and potential future research avenues.
- Open Access
- 26 June 2022
- Published by Association for Computing Machinery (ACM)
- p. 478-479
Documenting Web Data for Social Research (#DocuWeb22): A participatory workshop for developing structured and reusable practices
With this half-day, in-person workshop, we attempt to collaboratively discover best practices as well as frequent pitfalls encountered when working with Web data. Participants will first be presented with different perspectives on the significance of data quality in this specific context and familiarized with existing, structured approaches for the critical reflection on and documentation of data collection processes, before being invited to share their own experiences with the collection, use and documentation of Web data. We hope to thereby inspire participants to further integrate data documentation practices into their research processes, and for us to learn from the participants’ experiences in order to improve upon existing documentation frameworks for Web data. More details of the workshop, including the planned activities can be found at https://frohleon.github.io/DocuWeb22/.
- 26 June 2022
- Published by Association for Computing Machinery (ACM)
- p. 478-479
Open AccessDocumenting Web Data for Social Research (#DocuWeb22): A participatory workshop for developing structured and reusable practices
With this half-day, in-person workshop, we attempt to collaboratively discover best practices as well as frequent pitfalls encountered when working with Web data. Participants will first be presented with different perspectives on the significance of data quality in this specific context and familiarized with existing, structured approaches for the critical reflection on and documentation of data collection processes, before being invited to share their own experiences with the collection, use and documentation of Web data. We hope to thereby inspire participants to further integrate data documentation practices into their research processes, and for us to learn from the participants’ experiences in order to improve upon existing documentation frameworks for Web data. More details of the workshop, including the planned activities can be found at https://frohleon.github.io/DocuWeb22/.