Publications
Explore the publications from our RecSys laboratory.
1.
Neidhardt, Julia; Kolb, Thomas Elmar; Wagne, Ahmadou; Rippberger, Gwendolyn; Baeza-Yates, Ricardo
When Should Recommender Systems Not Act? Proceedings Article Forthcoming
In: Proceedings of the 20th ACM Conference on Recommender Systems (RecSys '26), Association for Computing Machinery, New York, NY, USA, Forthcoming.
@inproceedings{neidhardt2026when,
title = {When Should Recommender Systems Not Act?},
author = {Julia Neidhardt and Thomas Elmar Kolb and Ahmadou Wagne and Gwendolyn Rippberger and Ricardo Baeza-Yates},
year = {2026},
date = {2026-09-01},
urldate = {2026-09-01},
booktitle = {Proceedings of the 20th ACM Conference on Recommender Systems (RecSys '26)},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
keywords = {},
pubstate = {forthcoming},
tppubtype = {inproceedings}
}
2.
Kolb, Thomas Elmar; Starke, Alain Dominique; Trattner, Christoph
Exploring Revealed Behavior and Stated Profiles in Longitudinal News Recommendation Proceedings Article Forthcoming
In: Proceedings of the 20th ACM Conference on Recommender Systems (RecSys '26), Association for Computing Machinery, New York, NY, USA, Forthcoming.
@inproceedings{kolb2026exploring,
title = {Exploring Revealed Behavior and Stated Profiles in Longitudinal News Recommendation},
author = {Thomas Elmar Kolb and Alain Dominique Starke and Christoph Trattner},
year = {2026},
date = {2026-09-01},
urldate = {2026-09-01},
booktitle = {Proceedings of the 20th ACM Conference on Recommender Systems (RecSys '26)},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
keywords = {},
pubstate = {forthcoming},
tppubtype = {inproceedings}
}
3.
Modre, Laura; Wagne, Ahmadou; Kolb, Thomas Elmar; Neidhardt, Julia
Interaction Modality and Trust: Investigating System-Driven Conversational Recommendation and Faceted Search Proceedings Article Forthcoming
In: Proceedings of the 20th ACM Conference on Recommender Systems, Association for Computing Machinery, New York, NY, USA, Forthcoming.
@inproceedings{Modre2026,
title = {Interaction Modality and Trust: Investigating System-Driven Conversational Recommendation and Faceted Search},
author = {Laura Modre and Ahmadou Wagne and Thomas Elmar Kolb and Julia Neidhardt},
year = {2026},
date = {2026-08-24},
booktitle = {Proceedings of the 20th ACM Conference on Recommender Systems},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
keywords = {},
pubstate = {forthcoming},
tppubtype = {inproceedings}
}
4.
Wagne, Ahmadou; Kolb, Thomas; Banerjee, Ashmi; Nazary, Fatemeh; Neidhardt, Julia; Deldjoo, Yashar
Conversational Recommender Systems Using Generative Models (Gen-CRS): A Literature Review Journal Article
In: ACM Trans. Recomm. Syst., 2026, ISSN: 2770-6699.
@article{Wagne2026,
title = {Conversational Recommender Systems Using Generative Models (Gen-CRS): A Literature Review},
author = {Ahmadou Wagne and Thomas Kolb and Ashmi Banerjee and Fatemeh Nazary and Julia Neidhardt and Yashar Deldjoo},
doi = {10.1145/3828551},
issn = {2770-6699},
year = {2026},
date = {2026-07-07},
journal = {ACM Trans. Recomm. Syst.},
publisher = {Association for Computing Machinery (ACM)},
abstract = {Generative models are profoundly impacting how conversational recommender systems (CRS) are conceptualized, designed, and deployed in practice, enabling mixed-initiative, context-aware, and tool-augmented recommendation workflows that go beyond rigid traditional pipelines. This survey provides a comprehensive review of Generative Conversational Recommender Systems (Gen-CRS) from 2018–2025. We organize the field into three analytical layers: (i) architectural and methodological foundations, covering unified, modular, and agentic system designs, strategies for adapting foundation models, and methods for recommendation and response generation, (ii) knowledge and data foundations, detailing how item-level information (structured catalogs, knowledge graphs, unstructured and multi-modal content), user-level signals (short-term preferences, long-term profiles, and persona representations), and dialogue corpora and logs are integrated into generative workflows, and (iii) evaluation methodologies, structured around what is evaluated (output types), which quality dimensions are measured (e.g., task effectiveness, conversational quality, efficiency, user trust, and ethical concerns), how evaluation is conducted (offline metrics, simulation, user studies, online testing), who evaluates (humans, automated metrics, LLM-based judges), and which stakeholders are considered (consumers, item providers, platforms and society). Our analysis highlights both the capabilities and the risks introduced by generative components, including challenges in grounding, catalog fidelity, controllability, safety, and bias. We argue for responsible and transparent system design, emphasizing validated knowledge integration and evaluation protocols that reflect the full complexity of conversational interactions and system behaviors, and we outline open research challenges that must be addressed to develop reliable and trustworthy Gen-CRS. This survey is also informed by, and partly the result of, tutorial feedback we collected at ACM RecSys 2025 in Prague [42]. },
keywords = {},
pubstate = {published},
tppubtype = {article}
}
5.
Benson, Juliane; Zeh, Katharina; Essfors, Hannes; Fellner, Hannes; Neidhardt, Julia; Baumann, Andreas
Linguistic Diversity and Digitalization: An Ambivalent Relationship Proceedings Article
In: Hagedorn, Ludger; Schmid, Ute; Winter, Susan; Woltran, Stefan (Ed.): Digital Humanism, pp. 358–365, Springer Nature Switzerland, Cham, 2026, ISBN: 978-3-032-11108-1.
@inproceedings{10.1007/978-3-032-11108-1_26,
title = {Linguistic Diversity and Digitalization: An Ambivalent Relationship},
author = {Juliane Benson and Katharina Zeh and Hannes Essfors and Hannes Fellner and Julia Neidhardt and Andreas Baumann},
editor = {Ludger Hagedorn and Ute Schmid and Susan Winter and Stefan Woltran},
isbn = {978-3-032-11108-1},
year = {2026},
date = {2026-01-01},
booktitle = {Digital Humanism},
pages = {358–365},
publisher = {Springer Nature Switzerland},
address = {Cham},
abstract = {In this position paper, we argue that while digitalization amplifies biases towards only few languages dominating the linguistic landscape, modern language technology can help to mitigate language loss. We first elaborate on how the linguistic landscape in the digital and the non-digital sphere are distributionally different from each other in that the latter is strongly biased towards English, at the same time under-representing thousands of languages and the cultural knowledge that they encode. In a second step, we present results of qualitative interviews on individual linguistic experiences in the digital and the non-digital sphere that we have conducted in Québec, one of the provinces of Canada known for its linguistic diversity. These interviews highlight the potential that modern language technology have for safeguarding linguistic diversity. We conclude that the study of the impact of digitalization on the global linguistic landscape not only requires differential ways of measuring linguistic diversity but also a nuanced operationalization of digitalization.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
In this position paper, we argue that while digitalization amplifies biases towards only few languages dominating the linguistic landscape, modern language technology can help to mitigate language loss. We first elaborate on how the linguistic landscape in the digital and the non-digital sphere are distributionally different from each other in that the latter is strongly biased towards English, at the same time under-representing thousands of languages and the cultural knowledge that they encode. In a second step, we present results of qualitative interviews on individual linguistic experiences in the digital and the non-digital sphere that we have conducted in Québec, one of the provinces of Canada known for its linguistic diversity. These interviews highlight the potential that modern language technology have for safeguarding linguistic diversity. We conclude that the study of the impact of digitalization on the global linguistic landscape not only requires differential ways of measuring linguistic diversity but also a nuanced operationalization of digitalization.
6.
Wagne, Ahmadou
Investigating Preference Elicitation Strategies in Conversational Recommender Systems Proceedings Article
In: Proceedings of the 34th ACM Conference on User Modeling, Adaptation and Personalization, pp. 619–622, Association for Computing Machinery, New York, NY, USA, 2026, ISBN: 9798400723117.
@inproceedings{10.1145/3774935.3803059,
title = {Investigating Preference Elicitation Strategies in Conversational Recommender Systems},
author = {Ahmadou Wagne},
url = {https://doi.org/10.1145/3774935.3803059},
doi = {10.1145/3774935.3803059},
isbn = {9798400723117},
year = {2026},
date = {2026-01-01},
booktitle = {Proceedings of the 34th ACM Conference on User Modeling, Adaptation and Personalization},
pages = {619–622},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {UMAP '26},
abstract = {This research develops theoretical and methodological foundations for modular, mixed-initiative conversational recommender systems (CRS). It investigates how such systems can be conceptualized and systematically evaluated. The aim is to improve preference elicitation, dialogue management and intent detection by investigating the role of large language models (LLMs) in these tasks. First, a literature review is provided to conceptualize the use of generative models in CRS, and behavioral archetypes in user sessions that inform system design are identified. A user study compares a traditional faceted search interface with system-driven CRS to assess trust, usability and preference extraction quality in order to assess system acceptance. The outlined work extends this prototype to support mixed-initiative dialogues, with components for intent detection, question generation and state-tracking and focuses on a modular evaluation, often missing in the current literature. The project aims to deliver a holistic CRS designed and evaluated with consultancy of end users and platform stakeholders.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This research develops theoretical and methodological foundations for modular, mixed-initiative conversational recommender systems (CRS). It investigates how such systems can be conceptualized and systematically evaluated. The aim is to improve preference elicitation, dialogue management and intent detection by investigating the role of large language models (LLMs) in these tasks. First, a literature review is provided to conceptualize the use of generative models in CRS, and behavioral archetypes in user sessions that inform system design are identified. A user study compares a traditional faceted search interface with system-driven CRS to assess trust, usability and preference extraction quality in order to assess system acceptance. The outlined work extends this prototype to support mixed-initiative dialogues, with components for intent detection, question generation and state-tracking and focuses on a modular evaluation, often missing in the current literature. The project aims to deliver a holistic CRS designed and evaluated with consultancy of end users and platform stakeholders.
7.
Nalis, Irina; Klüwer, Nils; Neidhardt, Julia
From Labels to Context: Context-Sensitive User Modeling of Affect with Large Language Models Proceedings Article
In: Proceedings of the 34th ACM Conference on User Modeling, Adaptation and Personalization, pp. 490–493, Association for Computing Machinery, New York, NY, USA, 2026, ISBN: 9798400723117.
@inproceedings{10.1145/3774935.3812706,
title = {From Labels to Context: Context-Sensitive User Modeling of Affect with Large Language Models},
author = {Irina Nalis and Nils Klüwer and Julia Neidhardt},
url = {https://doi.org/10.1145/3774935.3812706},
doi = {10.1145/3774935.3812706},
isbn = {9798400723117},
year = {2026},
date = {2026-01-01},
booktitle = {Proceedings of the 34th ACM Conference on User Modeling, Adaptation and Personalization},
pages = {490–493},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {UMAP '26},
abstract = {Emotion-aware user modeling is increasingly important in adaptive systems, yet many approaches still represent affect as a fixed label inferred from isolated inputs. This limits their ability to capture how emotional meaning develops across context, time, and interaction history. We address this limitation by modeling affect as a user-level, context-dependent phenomenon. We introduce the context sphere, a structured user representation that aggregates interaction history and preserves conversational and temporal context. Building on this representation, we use a guided two-stage Large Language Model (LLM) pipeline to generate affective interpretations over extended interaction context. We evaluate the approach using automated consistency checks and an exploratory qualitative study with five semi-structured interviews. Results provide initial evidence of plausible, interpretable context-sensitive affective interpretation.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Emotion-aware user modeling is increasingly important in adaptive systems, yet many approaches still represent affect as a fixed label inferred from isolated inputs. This limits their ability to capture how emotional meaning develops across context, time, and interaction history. We address this limitation by modeling affect as a user-level, context-dependent phenomenon. We introduce the context sphere, a structured user representation that aggregates interaction history and preserves conversational and temporal context. Building on this representation, we use a guided two-stage Large Language Model (LLM) pipeline to generate affective interpretations over extended interaction context. We evaluate the approach using automated consistency checks and an exploratory qualitative study with five semi-structured interviews. Results provide initial evidence of plausible, interpretable context-sensitive affective interpretation.
8.
Wagne, Ahmadou; Kolb, Thomas Elmar; Banerjee, Ashmi; Neidhardt, Julia; Deldjoo, Yashar
LLM4Good: The 2nd Workshop on Sustainable and Trustworthy Large Language Models for Personalization Proceedings Article
In: Proceedings of the 34th ACM Conference on User Modeling, Adaptation and Personalization, pp. 684–687, Association for Computing Machinery, New York, NY, USA, 2026, ISBN: 9798400723117.
@inproceedings{10.1145/3774935.3802536,
title = {LLM4Good: The 2nd Workshop on Sustainable and Trustworthy Large Language Models for Personalization},
author = {Ahmadou Wagne and Thomas Elmar Kolb and Ashmi Banerjee and Julia Neidhardt and Yashar Deldjoo},
url = {https://doi.org/10.1145/3774935.3802536},
doi = {10.1145/3774935.3802536},
isbn = {9798400723117},
year = {2026},
date = {2026-01-01},
booktitle = {Proceedings of the 34th ACM Conference on User Modeling, Adaptation and Personalization},
pages = {684–687},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {UMAP '26},
abstract = {Large Language Models (LLMs) are transforming personalized services by enabling adaptive, context-aware recommendations and interactions. However, deploying these models at scale raises significant concerns about environmental impact, fairness, privacy, and trustworthiness, including high energy consumption, biased outputs, privacy breaches, and hallucinations. The LLM4Good workshop was already hosted at UMAP’251 and is a half-day workshop that addresses these challenges by fostering dialogue on sustainable and ethical approaches to LLM-based personalization. Participants will explore energy-efficient techniques, bias mitigation, privacy-preserving methods, and responsible deployment strategies. The workshop aligns with Sustainable Development Goals and Digital Humanism principles. It aims to guide the development of trustworthy, human-centric LLM systems that positively impact education, healthcare, and other domains.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Large Language Models (LLMs) are transforming personalized services by enabling adaptive, context-aware recommendations and interactions. However, deploying these models at scale raises significant concerns about environmental impact, fairness, privacy, and trustworthiness, including high energy consumption, biased outputs, privacy breaches, and hallucinations. The LLM4Good workshop was already hosted at UMAP’251 and is a half-day workshop that addresses these challenges by fostering dialogue on sustainable and ethical approaches to LLM-based personalization. Participants will explore energy-efficient techniques, bias mitigation, privacy-preserving methods, and responsible deployment strategies. The workshop aligns with Sustainable Development Goals and Digital Humanism principles. It aims to guide the development of trustworthy, human-centric LLM systems that positively impact education, healthcare, and other domains.
9.
Sergaš, Uroš; Wagne, Ahmadou; Kolb, Thomas Elmar; Neidhardt, Julia; Ferwerda, Bruce; Tkalcic, Marko
Prompt to Press: Evaluating Human Perception of AI Involvement in News Writing Across Prompt Specificity Proceedings Article
In: Companion Proceedings of the 31st International Conference on Intelligent User Interfaces, pp. 89–92, Association for Computing Machinery, New York, NY, USA, 2026, ISBN: 9798400719851.
@inproceedings{10.1145/3742414.3795097,
title = {Prompt to Press: Evaluating Human Perception of AI Involvement in News Writing Across Prompt Specificity},
author = {Uroš Sergaš and Ahmadou Wagne and Thomas Elmar Kolb and Julia Neidhardt and Bruce Ferwerda and Marko Tkalcic},
url = {https://doi.org/10.1145/3742414.3795097},
doi = {10.1145/3742414.3795097},
isbn = {9798400719851},
year = {2026},
date = {2026-01-01},
booktitle = {Companion Proceedings of the 31st International Conference on Intelligent User Interfaces},
pages = {89–92},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {IUI '26 Companion},
abstract = {Large language models (LLMs) are becoming a common feature in content creation tools, prompting important questions about how design choices influence user trust and engagement in AI-assisted journalism. Beyond output quality, factors such as prompt specificity, model choice, and authorship disclosure are themselves interaction design parameters that influence how users interpret and evaluate AI contributions. Yet, little is known about how these design decisions affect reader perceptions in journalistic contexts. To address this gap, we conducted an experiment with 150 participants who evaluated news articles on the sensitive topic of assisted suicide. The articles systematically varied in authorship (human-written, AI-edited, or AI-generated), stance (pro- or anti-legalization), and prompt specificity (vague, moderate, or highly detailed). Participants rated each article on engagement, subjectivity, and perceived AI involvement, and also provided open-ended justifications for their authorship judgments. Our findings show that prompt specificity and model choice significantly influence perceptions of authorship, underscoring how technical design decisions in AI tools can shape public trust in journalism.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Large language models (LLMs) are becoming a common feature in content creation tools, prompting important questions about how design choices influence user trust and engagement in AI-assisted journalism. Beyond output quality, factors such as prompt specificity, model choice, and authorship disclosure are themselves interaction design parameters that influence how users interpret and evaluate AI contributions. Yet, little is known about how these design decisions affect reader perceptions in journalistic contexts. To address this gap, we conducted an experiment with 150 participants who evaluated news articles on the sensitive topic of assisted suicide. The articles systematically varied in authorship (human-written, AI-edited, or AI-generated), stance (pro- or anti-legalization), and prompt specificity (vague, moderate, or highly detailed). Participants rated each article on engagement, subjectivity, and perceived AI involvement, and also provided open-ended justifications for their authorship judgments. Our findings show that prompt specificity and model choice significantly influence perceptions of authorship, underscoring how technical design decisions in AI tools can shape public trust in journalism.
10.
Pachinger, Pia; Goldzycher, Janis; Planitzer, Anna M.; Neidhardt, Julia; Hanbury, Allan
A Disaggregated Dataset on English Offensiveness Containing Spans Proceedings Article
In: Abercrombie, Gavin; Basile, Valerio; Frenda, Simona; Tonelli, Sara; Dudy, Shiran (Ed.): Proceedings of the The 4th Workshop on Perspectivist Approaches to NLP, pp. 1–14, Association for Computational Linguistics, Suzhou, China, 2025, ISBN: 979-8-89176-350-0.
@inproceedings{pachinger-etal-2025-disaggregated,
title = {A Disaggregated Dataset on English Offensiveness Containing Spans},
author = {Pia Pachinger and Janis Goldzycher and Anna M. Planitzer and Julia Neidhardt and Allan Hanbury},
editor = {Gavin Abercrombie and Valerio Basile and Simona Frenda and Sara Tonelli and Shiran Dudy},
url = {https://aclanthology.org/2025.nlperspectives-1.1/},
doi = {10.18653/v1/2025.nlperspectives-1.1},
isbn = {979-8-89176-350-0},
year = {2025},
date = {2025-11-01},
booktitle = {Proceedings of the The 4th Workshop on Perspectivist Approaches to NLP},
pages = {1–14},
publisher = {Association for Computational Linguistics},
address = {Suzhou, China},
abstract = {Toxicity labels at sub-document granularity and disaggregated labels lead to more nuanced and personalized toxicity classification and facilitate analysis. We re-annotate a subset of 1983 posts of the Jigsaw Toxic Comment Classification Challenge and provide disaggregated toxicity labels and spans that identify inappropriate language and targets of toxic statements. Manual analysis shows that five annotations per instance effectively capture meaningful disagreement patterns and allow for finer distinctions between genuine disagreement and that arising from annotation error or inconsistency. Our main findings are: (1) Disagreement often stems from divergent interpretations of edge-case toxicity (2) Disagreement is especially high in cases of toxic statements involving non-human targets (3) Disagreement on whether a passage consists of inappropriate language occurs not only on inherently questionable terms, but also on words that may be inappropriate in specific contexts while remaining acceptable in others (4) Transformer-based models effectively learn from aggregated data that reduces false negative classifications by being more sensitive towards minority opinions for posts to be toxic. We publish the new annotations under the CC BY 4.0 license.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Toxicity labels at sub-document granularity and disaggregated labels lead to more nuanced and personalized toxicity classification and facilitate analysis. We re-annotate a subset of 1983 posts of the Jigsaw Toxic Comment Classification Challenge and provide disaggregated toxicity labels and spans that identify inappropriate language and targets of toxic statements. Manual analysis shows that five annotations per instance effectively capture meaningful disagreement patterns and allow for finer distinctions between genuine disagreement and that arising from annotation error or inconsistency. Our main findings are: (1) Disagreement often stems from divergent interpretations of edge-case toxicity (2) Disagreement is especially high in cases of toxic statements involving non-human targets (3) Disagreement on whether a passage consists of inappropriate language occurs not only on inherently questionable terms, but also on words that may be inappropriate in specific contexts while remaining acceptable in others (4) Transformer-based models effectively learn from aggregated data that reduces false negative classifications by being more sensitive towards minority opinions for posts to be toxic. We publish the new annotations under the CC BY 4.0 license.
11.
Wagne, Ahmadou; Foll, Elen Le; Frantz, Florentine; Lasser, Jana
Giving the outrage a name – how researchers are challenging employment conditions under the hashtags #IchBinHanna and #IchBinReyhan Journal Article
In: Information, Communication & Society, pp. 1-27, 2025.
@article{Wagne_2024,
title = {Giving the outrage a name – how researchers are challenging employment conditions under the hashtags #IchBinHanna and #IchBinReyhan},
author = {Ahmadou Wagne and Elen Le Foll and Florentine Frantz and Jana Lasser},
editor = {Dan Mercea},
doi = {https://doi.org/10.1080/1369118X.2025.2452273},
year = {2025},
date = {2025-01-20},
urldate = {2025-01-20},
journal = {Information, Communication & Society},
pages = {1-27},
abstract = {Outraged by the release of a ministerial video in which short-term employment contracts in German academia were lauded through the embodiment of fictitious doctoral researcher Hanna, thousands of researchers rallied behind the hashtags #IchBinHanna & #IchBinReyhan to vent their frustrations about precarious academic employment in Germany. The emerging connective action attracted a comparatively large number of researchers in a short period of time, stayed active for over two years, elicited reactions from policymakers and the media, and influenced current legislative developments in Germany. We analyse the discourse of over 45,000 tweets related to the movement from its onset in June 2021 to March 2023. Using a mixed-methods approach that combines machine-learning, corpus-linguistic, and qualitative analysis methods, we aim to distil the factors that led to the movement’s considerable success. The fast growth of the movement was likely driven by the use of an easy-to-personalise action frame and the large variety of discussion topics, facilitating the involvement of different groups of academics across career levels and employment conditions. Our analysis of the linguistic characteristics of the discourse reveals largely positive, constructive, and active exchanges, in which many of the most salient keywords are lexical verbs. Our analyses offer an explanation for the continued involvement of many activists and the successful translation of solutions developed by the movement to news reports and proposed law amendments, despite an absence of coordination by any formal organisation.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Outraged by the release of a ministerial video in which short-term employment contracts in German academia were lauded through the embodiment of fictitious doctoral researcher Hanna, thousands of researchers rallied behind the hashtags #IchBinHanna & #IchBinReyhan to vent their frustrations about precarious academic employment in Germany. The emerging connective action attracted a comparatively large number of researchers in a short period of time, stayed active for over two years, elicited reactions from policymakers and the media, and influenced current legislative developments in Germany. We analyse the discourse of over 45,000 tweets related to the movement from its onset in June 2021 to March 2023. Using a mixed-methods approach that combines machine-learning, corpus-linguistic, and qualitative analysis methods, we aim to distil the factors that led to the movement’s considerable success. The fast growth of the movement was likely driven by the use of an easy-to-personalise action frame and the large variety of discussion topics, facilitating the involvement of different groups of academics across career levels and employment conditions. Our analysis of the linguistic characteristics of the discourse reveals largely positive, constructive, and active exchanges, in which many of the most salient keywords are lexical verbs. Our analyses offer an explanation for the continued involvement of many activists and the successful translation of solutions developed by the movement to news reports and proposed law amendments, despite an absence of coordination by any formal organisation.
12.
Pachinger, Pia; Planitzer, Anna Maria; Lecheler, Sophie; Neidhardt, Julia; Hanbury, Allan; Wegener, Rebekah
A Perspectivist Approach to Content Moderation: Incorporating User Perceptions of Online Norm Violations in Toxicity Detection Models Miscellaneous
2025.
@misc{ 20.500.12708_224144,
title = {A Perspectivist Approach to Content Moderation: Incorporating User Perceptions of Online Norm Violations in Toxicity Detection Models},
author = {Pia Pachinger and Anna Maria Planitzer and Sophie Lecheler and Julia Neidhardt and Allan Hanbury and Rebekah Wegener},
year = {2025},
date = {2025-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
13.
Pachinger, Pia; Planitzer, Anna Maria; Neidhardt, Julia; Hanbury, Allan; Lecheler, Sophie
Alignment by Disagreement? Toward Investigating LLMs’ Adaptation to Personal and Sociodemographic Variability in the Perception of Toxicity Miscellaneous
2025.
@misc{ 20.500.12708_224482,
title = {Alignment by Disagreement? Toward Investigating LLMs’ Adaptation to Personal and Sociodemographic Variability in the Perception of Toxicity},
author = {Pia Pachinger and Anna Maria Planitzer and Julia Neidhardt and Allan Hanbury and Sophie Lecheler},
year = {2025},
date = {2025-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
14.
Klüwer, Nils; Nalis, Irina; Neidhardt, Julia
Context over Categories: Implementing the Theory of Constructed Emotion with LLM-Guided User Analysis Proceedings Article
In: Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems, Association for Computing Machinery, New York, NY, USA, 2025, ISBN: 9798400713958.
@inproceedings{10.1145/3706599.3721205,
title = {Context over Categories: Implementing the Theory of Constructed Emotion with LLM-Guided User Analysis},
author = {Nils Klüwer and Irina Nalis and Julia Neidhardt},
url = {https://doi.org/10.1145/3706599.3721205},
doi = {10.1145/3706599.3721205},
isbn = {9798400713958},
year = {2025},
date = {2025-01-01},
booktitle = {Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {CHI EA '25},
abstract = {Emotion analysis is a critical research area with applications in content moderation and personalized systems. Many existing approaches rely on Ekman’s universal emotions theory, which reduces emotions to static categories, neglecting their complexity and contextual variability. This work introduces a novel, context-aware approach based on Lisa Feldman Barrett’s Theory of Constructed Emotion. A key contribution is the development of the “context sphere,” a personalized construct derived from user behavior data. To our knowledge, this is the first operationalization for computational methods. A context-aware emotion analysis pipeline was developed, incorporating advanced Large Language Model (LLM) prompting strategies like role-play and controlled generation. A case study in content moderation demonstrates how the “context sphere” enables contextually aware emotion analyses. Future directions include refining the framework, advancing LLM methodologies, and conducting user studies. This research lays the foundation for more human-centered, ethical, and effective emotion analysis systems.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Emotion analysis is a critical research area with applications in content moderation and personalized systems. Many existing approaches rely on Ekman’s universal emotions theory, which reduces emotions to static categories, neglecting their complexity and contextual variability. This work introduces a novel, context-aware approach based on Lisa Feldman Barrett’s Theory of Constructed Emotion. A key contribution is the development of the “context sphere,” a personalized construct derived from user behavior data. To our knowledge, this is the first operationalization for computational methods. A context-aware emotion analysis pipeline was developed, incorporating advanced Large Language Model (LLM) prompting strategies like role-play and controlled generation. A case study in content moderation demonstrates how the “context sphere” enables contextually aware emotion analyses. Future directions include refining the framework, advancing LLM methodologies, and conducting user studies. This research lays the foundation for more human-centered, ethical, and effective emotion analysis systems.
15.
Burke, Robin; Adomavicius, Gediminas; Bogers, Toine; Noia, Tommaso Di; Kowald, Dominik; Neidhardt, Julia; Özgöbek, Özlem; Pera, Maria Soledad; Tintarev, Nava; Ziegler, Jürgen
De-centering the (Traditional) user: Multistakeholder evaluation of recommender systems Journal Article
In: International Journal of Human-Computer Studies, vol. 203, pp. 103560, 2025, ISSN: 1071-5819.
@article{BURKE2025103560,
title = {De-centering the (Traditional) user: Multistakeholder evaluation of recommender systems},
author = {Robin Burke and Gediminas Adomavicius and Toine Bogers and Tommaso Di Noia and Dominik Kowald and Julia Neidhardt and Özlem Özgöbek and Maria Soledad Pera and Nava Tintarev and Jürgen Ziegler},
url = {https://www.sciencedirect.com/science/article/pii/S107158192500117X},
doi = {https://doi.org/10.1016/j.ijhcs.2025.103560},
issn = {1071-5819},
year = {2025},
date = {2025-01-01},
journal = {International Journal of Human-Computer Studies},
volume = {203},
pages = {103560},
abstract = {Multistakeholder recommender systems are those that account for the impacts and preferences of multiple groups of individuals, not just the end users receiving recommendations. Due to their complexity, these systems cannot be evaluated strictly by the overall utility of a single stakeholder, as is often the case of more mainstream recommender system applications. In this article, we focus our discussion on the challenges of multistakeholder evaluation of recommender systems. We bring attention to the different aspects involved—from the range of stakeholders involved (including but not limited to providers and consumers) to the values and specific goals of each relevant stakeholder. We discuss how to move from theoretical principles to practical implementation, providing specific use case examples. Finally, we outline open research directions for the RecSys community to explore. We aim to provide guidance to researchers and practitioners about incorporating these complex and domain-dependent issues of evaluation in the course of designing, developing, and researching applications with multistakeholder aspects.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Multistakeholder recommender systems are those that account for the impacts and preferences of multiple groups of individuals, not just the end users receiving recommendations. Due to their complexity, these systems cannot be evaluated strictly by the overall utility of a single stakeholder, as is often the case of more mainstream recommender system applications. In this article, we focus our discussion on the challenges of multistakeholder evaluation of recommender systems. We bring attention to the different aspects involved—from the range of stakeholders involved (including but not limited to providers and consumers) to the values and specific goals of each relevant stakeholder. We discuss how to move from theoretical principles to practical implementation, providing specific use case examples. Finally, we outline open research directions for the RecSys community to explore. We aim to provide guidance to researchers and practitioners about incorporating these complex and domain-dependent issues of evaluation in the course of designing, developing, and researching applications with multistakeholder aspects.
16.
Zamiechowska, Joanna; Neidhardt, Julia; Wörndl, Wolfgang; Kuflik, Tsvi; Livne, Amit; Zanker, Markus
CiRi-Engine: POI Recommender System for Diverse and Balanced Walking Tours Journal Article
In: 2025.
@article{ZamiechowskaJoanna2025CPRS,
title = {CiRi-Engine: POI Recommender System for Diverse and Balanced Walking Tours},
author = {Joanna Zamiechowska and Julia Neidhardt and Wolfgang Wörndl and Tsvi Kuflik and Amit Livne and Markus Zanker},
doi = {https://doi.org/10.34726/11839},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
publisher = {CEUR Workshop Proceedings},
address = {[Erscheinungsort nicht ermittelbar]},
edition = {Version of record},
abstract = {eng: We present CiRi-Engine (CityRiddler Recommendation Engine), an interactive city walking-tour recommender system. This demonstration paper showcases a novel approach to generating personalized and balanced itineraries for urban exploration. By combining user-specified constraints, such as start and end locations, tour duration, interest categories, and challenge preferences, with an efficient dual-stage routing algorithm, CiRi-Engine dynamically constructs diverse routes featuring curated Points of Interest (POIs). The engine leverages a novel hybrid of A* and Beam Search for path planning, and incorporates preference-aware POI selection to ensure both relevance and diversity. We demonstrate firsthand how the system balances route diversity, thematic coherence, and user-specified constraints, demonstrating its effectiveness for handling multiple objectives and generating engaging walking tours.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
eng: We present CiRi-Engine (CityRiddler Recommendation Engine), an interactive city walking-tour recommender system. This demonstration paper showcases a novel approach to generating personalized and balanced itineraries for urban exploration. By combining user-specified constraints, such as start and end locations, tour duration, interest categories, and challenge preferences, with an efficient dual-stage routing algorithm, CiRi-Engine dynamically constructs diverse routes featuring curated Points of Interest (POIs). The engine leverages a novel hybrid of A* and Beam Search for path planning, and incorporates preference-aware POI selection to ensure both relevance and diversity. We demonstrate firsthand how the system balances route diversity, thematic coherence, and user-specified constraints, demonstrating its effectiveness for handling multiple objectives and generating engaging walking tours.
17.
Neidhardt, Julia; Kuflik, Tsvi; Livne, Amit; Zanker, Markus; Wörndl, Wolfgang
Workshop on Recommenders in Tourism (RecTour) 2025 Proceedings Article
In: Proceedings of the Nineteenth ACM Conference on Recommender Systems, pp. 1412–1413, Association for Computing Machinery, New York, NY, USA, 2025, ISBN: 9798400713644.
@inproceedings{10.1145/3705328.3748500,
title = {Workshop on Recommenders in Tourism (RecTour) 2025},
author = {Julia Neidhardt and Tsvi Kuflik and Amit Livne and Markus Zanker and Wolfgang Wörndl},
url = {https://doi.org/10.1145/3705328.3748500},
doi = {10.1145/3705328.3748500},
isbn = {9798400713644},
year = {2025},
date = {2025-01-01},
booktitle = {Proceedings of the Nineteenth ACM Conference on Recommender Systems},
pages = {1412–1413},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {RecSys '25},
abstract = {The Workshop on Recommenders in Tourism (RecTour) has been successfully held in conjunction with the ACM Conference on Recommender Systems (RecSys) since 2016, with the exception of one year. This workshop focuses on the unique and evolving challenges of recommender systems in the tourism domain. Over time, RecTour has fostered an active community supported by both academia and industry. In this overview paper, we outline our motivations for organizing the RecTour workshop and highlight the main topics covered in RecTour submissions, including destination recommendation, privacy concerns in travel recommender systems, the cold-start problem, transformer-based approaches in recommendation systems, and best practices for evaluation and experimentation.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
The Workshop on Recommenders in Tourism (RecTour) has been successfully held in conjunction with the ACM Conference on Recommender Systems (RecSys) since 2016, with the exception of one year. This workshop focuses on the unique and evolving challenges of recommender systems in the tourism domain. Over time, RecTour has fostered an active community supported by both academia and industry. In this overview paper, we outline our motivations for organizing the RecTour workshop and highlight the main topics covered in RecTour submissions, including destination recommendation, privacy concerns in travel recommender systems, the cold-start problem, transformer-based approaches in recommendation systems, and best practices for evaluation and experimentation.
18.
Delić, Anđela; Ali, Syed Juned; Verbruggen, Charlotte; Neidhardt, Julia; Bork, Dominik
A Model Cleansing Pipeline for Model-Driven Engineering: Mitigating the Garbage In, Garbage Out Problem for Open Model Repositories Proceedings Article
In: 2025 ACM/IEEE 28th International Conference on Model Driven Engineering Languages and Systems (MODELS), pp. 60-71, 2025.
@inproceedings{11245421,
title = {A Model Cleansing Pipeline for Model-Driven Engineering: Mitigating the Garbage In, Garbage Out Problem for Open Model Repositories},
author = {Anđela Delić and Syed Juned Ali and Charlotte Verbruggen and Julia Neidhardt and Dominik Bork},
doi = {10.1109/MODELS67397.2025.00012},
year = {2025},
date = {2025-01-01},
booktitle = {2025 ACM/IEEE 28th International Conference on Model Driven Engineering Languages and Systems (MODELS)},
pages = {60-71},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
19.
Rippberger, Gwendolyn; Neidhardt, Julia
Comparative Analysis of Fashion Captioning for Multimodal Fashion Recommendation Proceedings Article
In: Pomo, Claudio; Jannach, Dietmar; Kim, Yubin; Malitesta, Daniele; Mancino, Alberto Carlo Maria; McAuley, Julian J.; Melchiorre, Alessandro B.; Nawaz, Shah (Ed.): Proceedings of the DaQuaMRec 2025 Workshop on Data Quality-Aware Multimodal Recommendation co-located with RecSys 2025, Prague, Czech Republic, September 22, 2025, pp. 8–19, CEUR-WS.org, 2025.
@inproceedings{DBLP:conf/daquamrec/RippbergerN25,
title = {Comparative Analysis of Fashion Captioning for Multimodal Fashion
Recommendation},
author = {Gwendolyn Rippberger and Julia Neidhardt},
editor = {Claudio Pomo and Dietmar Jannach and Yubin Kim and Daniele Malitesta and Alberto Carlo Maria Mancino and Julian J. McAuley and Alessandro B. Melchiorre and Shah Nawaz},
url = {https://ceur-ws.org/Vol-4188/paper1.pdf},
year = {2025},
date = {2025-01-01},
booktitle = {Proceedings of the DaQuaMRec 2025 Workshop on Data Quality-Aware Multimodal
Recommendation co-located with RecSys 2025, Prague, Czech Republic,
September 22, 2025},
pages = {8–19},
publisher = {CEUR-WS.org},
series = {CEUR Workshop Proceedings},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
20.
Kolb, Thomas Elmar; Banerjee, Ashmi; Wagne, Ahmadou; Neidhardt, Julia; Deldjoo, Yashar
LLM4Good: The 1st Workshop on Sustainable and Trustworthy Large Language Models for Personalization Proceedings Article
In: Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization, pp. 385–387, Association for Computing Machinery, New York, NY, USA, 2025, ISBN: 9798400713996.
@inproceedings{10.1145/3708319.3727560,
title = {LLM4Good: The 1st Workshop on Sustainable and Trustworthy Large Language Models for Personalization},
author = {Thomas Elmar Kolb and Ashmi Banerjee and Ahmadou Wagne and Julia Neidhardt and Yashar Deldjoo},
url = {https://doi.org/10.1145/3708319.3727560},
doi = {10.1145/3708319.3727560},
isbn = {9798400713996},
year = {2025},
date = {2025-01-01},
booktitle = {Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {385–387},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {UMAP Adjunct '25},
abstract = {Large Language Models (LLMs) are transforming personalized services by enabling adaptive, context-aware recommendations and interactions. However, deploying these models at scale raises significant concerns about environmental impact, fairness, privacy, and trustworthiness, including high energy consumption, biased outputs, privacy breaches, and hallucinations. The LLM4Good workshop is a half-day workshop that addresses these challenges by fostering dialogue on sustainable and ethical approaches to LLM-based personalization. Participants will explore energy-efficient techniques, bias mitigation, privacy-preserving methods, and responsible deployment strategies. The workshop aligns with Sustainable Development Goals and Digital Humanism principles. It aims to guide the development of trustworthy, human-centric LLM systems that positively impact education, healthcare, and other domains.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Large Language Models (LLMs) are transforming personalized services by enabling adaptive, context-aware recommendations and interactions. However, deploying these models at scale raises significant concerns about environmental impact, fairness, privacy, and trustworthiness, including high energy consumption, biased outputs, privacy breaches, and hallucinations. The LLM4Good workshop is a half-day workshop that addresses these challenges by fostering dialogue on sustainable and ethical approaches to LLM-based personalization. Participants will explore energy-efficient techniques, bias mitigation, privacy-preserving methods, and responsible deployment strategies. The workshop aligns with Sustainable Development Goals and Digital Humanism principles. It aims to guide the development of trustworthy, human-centric LLM systems that positively impact education, healthcare, and other domains.