Publications
Explore the publications from our RecSys laboratory.
21.
Kolb, Thomas Elmar; Nalis, Irina; Neidhardt, Julia
Bridging Preferences: Multi-Stakeholder Insights on Ideal News Recommendations Proceedings Article
In: Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization, pp. 268–272, Association for Computing Machinery, New York, NY, USA, 2025, ISBN: 9798400713132.
@inproceedings{10.1145/3699682.3728355,
title = {Bridging Preferences: Multi-Stakeholder Insights on Ideal News Recommendations},
author = {Thomas Elmar Kolb and Irina Nalis and Julia Neidhardt},
url = {https://doi.org/10.1145/3699682.3728355},
doi = {10.1145/3699682.3728355},
isbn = {9798400713132},
year = {2025},
date = {2025-01-01},
booktitle = {Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {268–272},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {UMAP '25},
abstract = {In the evolving realm of recommender systems, our study contributes to the understanding of potential improvements in news recommendation beyond accuracy. Central to our research is the integration of insights from news industry experts and prospective readers, compared with automated news recommendations. We conducted a labeling study with 168 articles, using Best-Worst Scaling (BWS) for ranking and topic modeling. This approach enabled a thorough examination of stakeholder expectations for ideal reading recommendations, specifically by investigating the gap between stated and revealed preferences. Our findings show alignment in ranking behavior among journalists, prospective readers, and the BM-25 algorithm. However, preferences for different beyond-accuracy measures varied. Accompanying this work, a corpus of news articles and the labeled rankings have been made available.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
In the evolving realm of recommender systems, our study contributes to the understanding of potential improvements in news recommendation beyond accuracy. Central to our research is the integration of insights from news industry experts and prospective readers, compared with automated news recommendations. We conducted a labeling study with 168 articles, using Best-Worst Scaling (BWS) for ranking and topic modeling. This approach enabled a thorough examination of stakeholder expectations for ideal reading recommendations, specifically by investigating the gap between stated and revealed preferences. Our findings show alignment in ranking behavior among journalists, prospective readers, and the BM-25 algorithm. However, preferences for different beyond-accuracy measures varied. Accompanying this work, a corpus of news articles and the labeled rankings have been made available.
22.
Kolb, Thomas Elmar; Wagne, Ahmadou; Banerjee, Ashmi; Nazary, Fatemeh; Neidhardt, Julia; Deldjoo, Yashar; Noia, Tommaso Di
A Tutorial on Recent Advances in Generative Conversational Recommender Systems Proceedings Article
In: Proceedings of the Nineteenth ACM Conference on Recommender Systems, pp. 1420–1422, Association for Computing Machinery, New York, NY, USA, 2025, ISBN: 9798400713644.
@inproceedings{10.1145/3705328.3748010,
title = {A Tutorial on Recent Advances in Generative Conversational Recommender Systems},
author = {Thomas Elmar Kolb and Ahmadou Wagne and Ashmi Banerjee and Fatemeh Nazary and Julia Neidhardt and Yashar Deldjoo and Tommaso Di Noia},
url = {https://doi.org/10.1145/3705328.3748010},
doi = {10.1145/3705328.3748010},
isbn = {9798400713644},
year = {2025},
date = {2025-01-01},
booktitle = {Proceedings of the Nineteenth ACM Conference on Recommender Systems},
pages = {1420–1422},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {RecSys '25},
abstract = {Conversational recommender systems (CRSs) are increasingly vital for delivering multi-turn, context-aware recommendations. This tutorial provides a concise yet comprehensive exploration of modern generative CRSs, highlighting recent advances in generative AI—such as breakthroughs in large language models and neural generation pipelines, that enhance dialogue management, user modeling, and response generation. In addition, the tutorial addresses core challenges, including data acquisition, multi-turn personalization, and evaluation issues, such as controlling hallucinations, accounting for social factors, and managing ethical considerations, while also discussing emerging risks and novel solutions. Ultimately, participants will be equipped with actionable insights and practical tools for building new conversational recommender systems powered by generative models.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Conversational recommender systems (CRSs) are increasingly vital for delivering multi-turn, context-aware recommendations. This tutorial provides a concise yet comprehensive exploration of modern generative CRSs, highlighting recent advances in generative AI—such as breakthroughs in large language models and neural generation pipelines, that enhance dialogue management, user modeling, and response generation. In addition, the tutorial addresses core challenges, including data acquisition, multi-turn personalization, and evaluation issues, such as controlling hallucinations, accounting for social factors, and managing ethical considerations, while also discussing emerging risks and novel solutions. Ultimately, participants will be equipped with actionable insights and practical tools for building new conversational recommender systems powered by generative models.
23.
Steindl, Bernhard; Kolb, Thomas Elmar; Neidhardt, Julia
Beyond Demographics: Evaluating News Recommender Systems Fairness Through Behavioural Communities Proceedings Article
In: Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization, pp. 13–17, Association for Computing Machinery, New York, NY, USA, 2025, ISBN: 9798400713996.
@inproceedings{10.1145/3708319.3733694,
title = {Beyond Demographics: Evaluating News Recommender Systems Fairness Through Behavioural Communities},
author = {Bernhard Steindl and Thomas Elmar Kolb and Julia Neidhardt},
url = {https://doi.org/10.1145/3708319.3733694},
doi = {10.1145/3708319.3733694},
isbn = {9798400713996},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
booktitle = {Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {13–17},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
series = {UMAP Adjunct '25},
abstract = {Fairness in recommender systems is often framed around demographic attributes. In this work, we explore a novel direction—evaluating fairness across latent behavioural communities derived from user interactions on a real-world news platform. Using graph-based community detection (Louvain and Infomap), we identify large user groups and examine how different network modelling choices affect fairness outcomes in both traditional and fairness-aware recommender systems. Experiments on an Austrian news dataset reveal that small changes in graph construction considerably impact community formation and recommendation quality. Notably, fairness-aware algorithms show only marginal improvements over standard approaches, underscoring the complexity of achieving equitable outcomes in real-world systems and raising important questions for future research.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Fairness in recommender systems is often framed around demographic attributes. In this work, we explore a novel direction—evaluating fairness across latent behavioural communities derived from user interactions on a real-world news platform. Using graph-based community detection (Louvain and Infomap), we identify large user groups and examine how different network modelling choices affect fairness outcomes in both traditional and fairness-aware recommender systems. Experiments on an Austrian news dataset reveal that small changes in graph construction considerably impact community formation and recommendation quality. Notably, fairness-aware algorithms show only marginal improvements over standard approaches, underscoring the complexity of achieving equitable outcomes in real-world systems and raising important questions for future research.
24.
Burke, Robin; Adomavicius, Gediminas; Bogers, Toine; Noia, Tomasso Di; Kowald, Dominik; Neidhardt, Julia; Özgöbek, Özlem; Pera, Maria; Ziegler, Jürgen
Multistakeholder and Multimethod Evaluation Book Chapter
In: Evaluation Perspectives of Recommender Systems: Driving Research and Education (Dagstuhl Seminar 24211), vol. 14, no. 5, pp. 123–145, Schloss Dagstuhl – Leibniz-Zentrum fuer Informatik GmbH, 5, 2024.
@inbook{b077f33cc308407bb4fe6a2ee2336dfe,
title = {Multistakeholder and Multimethod Evaluation},
author = {Robin Burke and Gediminas Adomavicius and Toine Bogers and Tomasso Di Noia and Dominik Kowald and Julia Neidhardt and Özlem Özgöbek and Maria Pera and Jürgen Ziegler},
year = {2024},
date = {2024-11-26},
booktitle = {Evaluation Perspectives of Recommender Systems: Driving Research and Education (Dagstuhl Seminar 24211)},
volume = {14},
number = {5},
pages = {123–145},
publisher = {Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik GmbH},
edition = {5},
series = {Dagstuhl Seminar Proceedings},
abstract = {Multistakeholder recommender systems are defined by [1] as those that account for “the preferences of multiple parties when generating recommendations, especially when these parties are on different sides of the recommendation interaction.” Due to their complexity, evaluating these systems cannot be restricted to the overall utility of a single stakeholder, as is often the case of more mainstream recommender system applications. In this section, we focus our discussion on the intricacies involved in understanding what is the “right” construct required to ensure the proper evaluation of multistakeholder recommender systems. We bring attention to the different aspects involved in the evaluation of multistakeholder recommender systems – from the range of stakeholders involved (beyond producers and consumers) to the values and specific goals of each relevant stakeholder. Additionally, we discuss how to move from theoretical evaluation to practical implementation, providing specific use case examples. Finally, we outline open research directions for the RecSys community to explore. Our aim in this section is to provide guidance to researchers and practitioners about how to think about these complex and domain-dependent issues in the course of designing, developing, and researching applications with multistakeholder aspects.},
keywords = {},
pubstate = {published},
tppubtype = {inbook}
}
Multistakeholder recommender systems are defined by [1] as those that account for “the preferences of multiple parties when generating recommendations, especially when these parties are on different sides of the recommendation interaction.” Due to their complexity, evaluating these systems cannot be restricted to the overall utility of a single stakeholder, as is often the case of more mainstream recommender system applications. In this section, we focus our discussion on the intricacies involved in understanding what is the “right” construct required to ensure the proper evaluation of multistakeholder recommender systems. We bring attention to the different aspects involved in the evaluation of multistakeholder recommender systems – from the range of stakeholders involved (beyond producers and consumers) to the values and specific goals of each relevant stakeholder. Additionally, we discuss how to move from theoretical evaluation to practical implementation, providing specific use case examples. Finally, we outline open research directions for the RecSys community to explore. Our aim in this section is to provide guidance to researchers and practitioners about how to think about these complex and domain-dependent issues in the course of designing, developing, and researching applications with multistakeholder aspects.
25.
Wagne, Ahmadou; Neidhardt, Julia
Can We Integrate Items into Models? Knowledge Editing to Align LLMs with Product Catalogs Proceedings Article
In: Anelli, Vito Walter; Basile, Pierpaolo; Noia, Tommaso Di; Donini, Francesco Maria; Ferrara, Antonio; Musto, Cataldo; Narducci, Fedelucio; Ragone, Azzurra; Zanker, Markus (Ed.): Sixth Knowledge-aware and Conversational Recommender Systems (KaRS) Workshop @ RecSys 2024, pp. 56-65, CEUR-WS.org, Bari, Italy, 2024.
@inproceedings{Wagne_Neidhardt_2024_2,
title = {Can We Integrate Items into Models? Knowledge Editing to Align LLMs with Product Catalogs},
author = {Ahmadou Wagne and Julia Neidhardt},
editor = {Vito Walter Anelli and Pierpaolo Basile and Tommaso Di Noia and Francesco Maria Donini and Antonio Ferrara and Cataldo Musto and Fedelucio Narducci and Azzurra Ragone and Markus Zanker},
year = {2024},
date = {2024-10-31},
urldate = {2024-10-31},
booktitle = {Sixth Knowledge-aware and Conversational Recommender Systems (KaRS) Workshop @ RecSys 2024},
volume = {3817},
pages = {56-65},
publisher = {CEUR-WS.org},
address = {Bari, Italy},
abstract = {This study explores the potential of knowledge editing techniques to enhance Large Language Models (LLMs)
for Conversational Recommender Systems (CRS). While LLMs like GPT, Llama, and Gemini have advanced
conversational capabilities, they face challenges in representing dynamic, real-world item catalogs, often leading to
inaccuracies and hallucinations in recommendations. This research preliminarily investigates whether knowledge
editing can address these limitations by updating the internal knowledge of LLMs, thereby improving the
accuracy of product information without full model retraining. Using the open-source Llama2 model, we apply
two knowledge editing methods (GRACE and r-ROME) on a dataset of notebook listings. Our findings demonstrate
improvements in the model’s ability to accurately represent product features, with r-ROME achieving the highest
gain, while not decreasing model efficiency. The study highlights the perspective of utilizing knowledge editing
to enhance CRS and suggests future work to explore broader applications and impacts on recommender systems
performance},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This study explores the potential of knowledge editing techniques to enhance Large Language Models (LLMs)
for Conversational Recommender Systems (CRS). While LLMs like GPT, Llama, and Gemini have advanced
conversational capabilities, they face challenges in representing dynamic, real-world item catalogs, often leading to
inaccuracies and hallucinations in recommendations. This research preliminarily investigates whether knowledge
editing can address these limitations by updating the internal knowledge of LLMs, thereby improving the
accuracy of product information without full model retraining. Using the open-source Llama2 model, we apply
two knowledge editing methods (GRACE and r-ROME) on a dataset of notebook listings. Our findings demonstrate
improvements in the model’s ability to accurately represent product features, with r-ROME achieving the highest
gain, while not decreasing model efficiency. The study highlights the perspective of utilizing knowledge editing
to enhance CRS and suggests future work to explore broader applications and impacts on recommender systems
performance
for Conversational Recommender Systems (CRS). While LLMs like GPT, Llama, and Gemini have advanced
conversational capabilities, they face challenges in representing dynamic, real-world item catalogs, often leading to
inaccuracies and hallucinations in recommendations. This research preliminarily investigates whether knowledge
editing can address these limitations by updating the internal knowledge of LLMs, thereby improving the
accuracy of product information without full model retraining. Using the open-source Llama2 model, we apply
two knowledge editing methods (GRACE and r-ROME) on a dataset of notebook listings. Our findings demonstrate
improvements in the model’s ability to accurately represent product features, with r-ROME achieving the highest
gain, while not decreasing model efficiency. The study highlights the perspective of utilizing knowledge editing
to enhance CRS and suggests future work to explore broader applications and impacts on recommender systems
performance
26.
Wagne, Ahmadou; Neidhardt, Julia
What to compare? Towards understanding user sessions on price comparison platforms Proceedings Article
In: Noia, Tommaso Di; Lops, Pasquale; Joachims, Thorsten; Verbert, Katrien; Castells, Pablo; Dong, Zhenhua; London, Ben (Ed.): RecSys '24: Proceedings of the 18th ACM Conference on Recommender Systems, pp. 1158 – 1162, Association for Computing Machinery, New York, NY, United States, 2024, ISBN: 979-8-4007-1127-5.
@inproceedings{Wagne_Neidhardt_2024,
title = {What to compare? Towards understanding user sessions on price comparison platforms},
author = {Ahmadou Wagne and Julia Neidhardt},
editor = {Tommaso Di Noia and Pasquale Lops and Thorsten Joachims and Katrien Verbert and Pablo Castells and Zhenhua Dong and Ben London},
doi = {https://doi.org/10.1145/3640457.3691717},
isbn = {979-8-4007-1127-5},
year = {2024},
date = {2024-10-08},
urldate = {2024-10-08},
booktitle = {RecSys '24: Proceedings of the 18th ACM Conference on Recommender Systems},
pages = {1158 - 1162},
publisher = {Association for Computing Machinery},
address = {New York, NY, United States},
abstract = {E-commerce and online shopping have become integral to the lives of many, with various user behavior types historically identified. Beyond deciding what to buy, determining where to make a purchase has led to the importance of price comparison platforms. However, user behavior on these platforms remains underexplored. Furthermore, web analytics often struggle with tracking users over time and deriving meaningful user types from data. This paper addresses these gaps by defining session types through the analysis and clustering of user logs from a major price comparison platform. The study identifies six distinct session clusters: quick peek, major purchase, constraint-based browsing, knowledge seeking, search and browse and heavy browsing. These findings are intended to inform the design and development of a conversational recommender system (CRS). Often, CRS development occurs without adequate consideration of the existing system into which it will be integrated. The study’s findings, derived from both quantitative analysis and expert interviews, provide valuable contributions, including identified session clusters, their interpretation and indicators on which users might benefit from a CRS on these platforms.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
E-commerce and online shopping have become integral to the lives of many, with various user behavior types historically identified. Beyond deciding what to buy, determining where to make a purchase has led to the importance of price comparison platforms. However, user behavior on these platforms remains underexplored. Furthermore, web analytics often struggle with tracking users over time and deriving meaningful user types from data. This paper addresses these gaps by defining session types through the analysis and clustering of user logs from a major price comparison platform. The study identifies six distinct session clusters: quick peek, major purchase, constraint-based browsing, knowledge seeking, search and browse and heavy browsing. These findings are intended to inform the design and development of a conversational recommender system (CRS). Often, CRS development occurs without adequate consideration of the existing system into which it will be integrated. The study’s findings, derived from both quantitative analysis and expert interviews, provide valuable contributions, including identified session clusters, their interpretation and indicators on which users might benefit from a CRS on these platforms.
27.
Pachinger, Pia; Goldzycher, Janis; Planitzer, Anna; Kusa, Wojciech; Hanbury, Allan; Neidhardt, Julia
ÄustroTox: A Dataset for Target-Based Austrian German Offensive Language Detection” Proceedings Article
In: Ku, Lun-Wei; Martins, Andre; Srikumar, Vivek (Ed.): Findings of the Association for Computational Linguistics: ACL 2024, pp. 11990–12001, Association for Computational Linguistics, Bangkok, Thailand, 2024.
@inproceedings{pachinger-etal-2024-austrotox,
title = {ÄustroTox: A Dataset for Target-Based Austrian German Offensive Language Detection"},
author = {Pia Pachinger and Janis Goldzycher and Anna Planitzer and Wojciech Kusa and Allan Hanbury and Julia Neidhardt},
editor = {Lun-Wei Ku and Andre Martins and Vivek Srikumar},
url = {https://aclanthology.org/2024.findings-acl.713/},
doi = {10.18653/v1/2024.findings-acl.713},
year = {2024},
date = {2024-08-01},
booktitle = {Findings of the Association for Computational Linguistics: ACL 2024},
pages = {11990–12001},
publisher = {Association for Computational Linguistics},
address = {Bangkok, Thailand},
abstract = {Model interpretability in toxicity detection greatly profits from token-level annotations. However, currently, such annotations are only available in English. We introduce a dataset annotated for offensive language detection sourced from a news forum, notable for its incorporation of the Austrian German dialect, comprising 4,562 user comments. In addition to binary offensiveness classification, we identify spans within each comment constituting vulgar language or representing targets of offensive statements. We evaluate fine-tuned Transformer models as well as large language models in a zero- and few-shot fashion. The results indicate that while fine-tuned models excel in detecting linguistic peculiarities such as vulgar dialect, large language models demonstrate superior performance in detecting offensiveness in AustroTox.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Model interpretability in toxicity detection greatly profits from token-level annotations. However, currently, such annotations are only available in English. We introduce a dataset annotated for offensive language detection sourced from a news forum, notable for its incorporation of the Austrian German dialect, comprising 4,562 user comments. In addition to binary offensiveness classification, we identify spans within each comment constituting vulgar language or representing targets of offensive statements. We evaluate fine-tuned Transformer models as well as large language models in a zero- and few-shot fashion. The results indicate that while fine-tuned models excel in detecting linguistic peculiarities such as vulgar dialect, large language models demonstrate superior performance in detecting offensiveness in AustroTox.
28.
Wagne, Ahmadou; Neidhardt, Julia; Kolb, Thomas Elmar
PopAut: An Annotated Corpus for Populism Detection in Austrian News Comments Proceedings Article
In: Calzolari, Nicoletta; Kan, Min-Yen; Hoste, Veronique; Lenci, Alessandro; Sakti, Sakriani; Xue, Nianwen (Ed.): Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pp. 12879–12892, ELRA and ICCL, Torino, Italy, 2024.
@inproceedings{Wagne_Neidhardt_Kolb_2024,
title = {PopAut: An Annotated Corpus for Populism Detection in Austrian News Comments},
author = {Ahmadou Wagne and Julia Neidhardt and Thomas Elmar Kolb},
editor = {Nicoletta Calzolari and Min-Yen Kan and Veronique Hoste and Alessandro Lenci and Sakriani Sakti and Nianwen Xue},
url = {https://aclanthology.org/2024.lrec-main.1128},
year = {2024},
date = {2024-05-01},
urldate = {2024-05-01},
booktitle = {Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)},
pages = {12879–12892},
publisher = {ELRA and ICCL},
address = {Torino, Italy},
abstract = {Populism is a phenomenon that is noticeably present in the political landscape of various countries over the past decades. While populism expressed by politicians has been thoroughly examined in the literature, populism expressed by citizens is still underresearched, especially when it comes to its automated detection in text. This work presents the PopAut corpus, which is the first annotated corpus of news comments for populism in the German language. It features 1,200 comments collected between 2019-2021 that are annotated for populist motives anti-elitism, people-centrism and people-sovereignty. Following the definition of Cas Mudde, populism is seen as a thin ideology. This work shows that annotators reach a high agreement when labeling news comments for these motives. The data set is collected to serve as the basis for automated populism detection using machine-learning methods. By using transformer-based models, we can outperform existing dictionaries tailored for automated populism detection in German social media content. Therefore our work provides a rich resource for future work on the classification of populist user comments in the German language.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Populism is a phenomenon that is noticeably present in the political landscape of various countries over the past decades. While populism expressed by politicians has been thoroughly examined in the literature, populism expressed by citizens is still underresearched, especially when it comes to its automated detection in text. This work presents the PopAut corpus, which is the first annotated corpus of news comments for populism in the German language. It features 1,200 comments collected between 2019-2021 that are annotated for populist motives anti-elitism, people-centrism and people-sovereignty. Following the definition of Cas Mudde, populism is seen as a thin ideology. This work shows that annotators reach a high agreement when labeling news comments for these motives. The data set is collected to serve as the basis for automated populism detection using machine-learning methods. By using transformer-based models, we can outperform existing dictionaries tailored for automated populism detection in German social media content. Therefore our work provides a rich resource for future work on the classification of populist user comments in the German language.
29.
Pachinger, Pia; Goldzycher, Janis; Planitzer, Anna Maria; Kusa, Wojciech; Hanbury, Allan; Neidhardt, Julia
A Dataset for Span-Based Austrian German and English Offensive Language Detection Miscellaneous
2024.
@misc{ 20.500.12708_210352,
title = {A Dataset for Span-Based Austrian German and English Offensive Language Detection},
author = {Pia Pachinger and Janis Goldzycher and Anna Maria Planitzer and Wojciech Kusa and Allan Hanbury and Julia Neidhardt},
year = {2024},
date = {2024-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
30.
Nalis, Irina; Neidhardt, Julia
Towards Possibility: Interdisciplinary Perspectives on Enhancing Recommender Systems Beyond Accuracy Miscellaneous
2024.
@misc{ 20.500.12708_210789,
title = {Towards Possibility: Interdisciplinary Perspectives on Enhancing Recommender Systems Beyond Accuracy},
author = {Irina Nalis and Julia Neidhardt},
year = {2024},
date = {2024-01-01},
abstract = {In this introductory lecture, Dr. Julia Neidhardt, Director of the CDLab for Recommender Systems and UNESCO Co-Chair in Digital Humanism, and Dr. Irina Nalis, psychologist, and interdisciplinary researcher at the CDLab, explore recommender systems from a Digital Humanism viewpoint, focusing on the intersection of technology, psychology, and societal needs. Addressing the limitations and risks of accuracy-centric metrics, they emphasize the importance of establishing new research and development methods to go beyond accuracy and towards more human-potential centered recommendations. Their lecture, based on interdisciplinary research including advanced algorithms, cross-domain recommendations, and the integration of large-language models, demonstrates the potential of recommender systems to foster diversity, serendipity, and democratic fairness. They further discuss the role of choice architecture and affordances in making responsible recommendations, highlighting the importance of aligning with broader policies like the EU Digital Services Act to meet societal needs and enrich the digital landscape.},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
In this introductory lecture, Dr. Julia Neidhardt, Director of the CDLab for Recommender Systems and UNESCO Co-Chair in Digital Humanism, and Dr. Irina Nalis, psychologist, and interdisciplinary researcher at the CDLab, explore recommender systems from a Digital Humanism viewpoint, focusing on the intersection of technology, psychology, and societal needs. Addressing the limitations and risks of accuracy-centric metrics, they emphasize the importance of establishing new research and development methods to go beyond accuracy and towards more human-potential centered recommendations. Their lecture, based on interdisciplinary research including advanced algorithms, cross-domain recommendations, and the integration of large-language models, demonstrates the potential of recommender systems to foster diversity, serendipity, and democratic fairness. They further discuss the role of choice architecture and affordances in making responsible recommendations, highlighting the importance of aligning with broader policies like the EU Digital Services Act to meet societal needs and enrich the digital landscape.
31.
Neidhardt, Julia; Kuflik, Tsvi; Livne, Amit; Zanker, Markus
Workshop on Recommenders in Tourism (RecTour) 2024 Proceedings Article
In: Proceedings of the 18th ACM Conference on Recommender Systems, pp. 1229–1231, Association for Computing Machinery, Bari, Italy, 2024, ISBN: 9798400705052.
@inproceedings{10.1145/3640457.3687107,
title = {Workshop on Recommenders in Tourism (RecTour) 2024},
author = {Julia Neidhardt and Tsvi Kuflik and Amit Livne and Markus Zanker},
url = {https://doi.org/10.1145/3640457.3687107},
doi = {10.1145/3640457.3687107},
isbn = {9798400705052},
year = {2024},
date = {2024-01-01},
booktitle = {Proceedings of the 18th ACM Conference on Recommender Systems},
pages = {1229–1231},
publisher = {Association for Computing Machinery},
address = {Bari, Italy},
series = {RecSys '24},
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. This year, the workshop features a special challenge focused on ranking travel reviews. 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. This year, the workshop features a special challenge focused on ranking travel reviews. 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.
32.
Aayesha, Aayesha; Afzaal, Muhammad; Neidhardt, Julia
Social Circle-Enhanced Fashion Recommendations System Proceedings Article
In: Brusilovsky, Peter; Gemmis, Marco; Felfernig, Alexander (Ed.): pp. 81–91, 2024.
@inproceedings{ 20.500.12708_208019,
title = {Social Circle-Enhanced Fashion Recommendations System},
author = {Aayesha Aayesha and Muhammad Afzaal and Julia Neidhardt},
editor = {Peter Brusilovsky and Marco Gemmis and Alexander Felfernig},
year = {2024},
date = {2024-01-01},
volume = {3815},
pages = {81–91},
abstract = {When shopping for fashionable clothing items, consumers frequently experience indecision and struggle to make choices, resulting in a stalling of the purchasing process. In such scenarios, most often they need support of their friends from their social circle to choose suitable clothes for different events. To provide decision-making support, considerable research has focused on generating social-aware recommendations that incorporate input from the user’s social circle. However, there has been minimal research dedicated to develop and evaluate such systems that could assess the importance of social circles in producing social-aware fashion recommendations and identifying factors that might enhance these recommendations. This paper addresses these limitations by developing a Social Circle-Enhanced Fashion Recommendation (SCEFR) System that encompasses friends feedback to generate recommendations. The SCEFR system was evaluated by conducting a user study, comparing system-generated recommendations with user choices as rank correlation coefficients. The findings indicate that inputs from the social circle alone have limited potential in generating effective social-aware recommendations. However, when the user’s shopping preferences were shared with their social circle, the quality of these recommendations significantly improved, as evidenced by a qualitative analysis of user feedback. Furthermore, in comparative analysis with the state-of-the-art (SOTA) approaches of recommendation generation, the SCEFR system informed by user’s shopping preferences demonstrated superiority.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
When shopping for fashionable clothing items, consumers frequently experience indecision and struggle to make choices, resulting in a stalling of the purchasing process. In such scenarios, most often they need support of their friends from their social circle to choose suitable clothes for different events. To provide decision-making support, considerable research has focused on generating social-aware recommendations that incorporate input from the user’s social circle. However, there has been minimal research dedicated to develop and evaluate such systems that could assess the importance of social circles in producing social-aware fashion recommendations and identifying factors that might enhance these recommendations. This paper addresses these limitations by developing a Social Circle-Enhanced Fashion Recommendation (SCEFR) System that encompasses friends feedback to generate recommendations. The SCEFR system was evaluated by conducting a user study, comparing system-generated recommendations with user choices as rank correlation coefficients. The findings indicate that inputs from the social circle alone have limited potential in generating effective social-aware recommendations. However, when the user’s shopping preferences were shared with their social circle, the quality of these recommendations significantly improved, as evidenced by a qualitative analysis of user feedback. Furthermore, in comparative analysis with the state-of-the-art (SOTA) approaches of recommendation generation, the SCEFR system informed by user’s shopping preferences demonstrated superiority.
33.
Aayesha, Aayesha; Afzaal, Muhammad; Neidhardt, Julia
User Experience of Recommender System: A User Study of Social-aware Fashion Recommendations System Proceedings Article
In: Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization, pp. 356–361, Association for Computing Machinery, Cagliari, Italy, 2024, ISBN: 9798400704666.
@inproceedings{10.1145/3631700.3664896,
title = {User Experience of Recommender System: A User Study of Social-aware Fashion Recommendations System},
author = {Aayesha Aayesha and Muhammad Afzaal and Julia Neidhardt},
url = {https://doi.org/10.1145/3631700.3664896},
doi = {10.1145/3631700.3664896},
isbn = {9798400704666},
year = {2024},
date = {2024-01-01},
booktitle = {Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {356–361},
publisher = {Association for Computing Machinery},
address = {Cagliari, Italy},
series = {UMAP Adjunct '24},
abstract = {User experience, which encompasses users’ feelings and perceptions, is regarded as a key element in the evaluation of recommender systems. The existing literature extensively works on recommendation generation strategies with focus on the accuracy by considering objective aspects of the system. Although some of the current works considered subjective aspects of the recommendation systems from a user-centric perspective to evaluate the recommender system, however, a comprehensive analysis that could investigate factors to improve user experience was of limited focus. In this paper, we propose a methodology that provides a comprehensive multi-perspective analysis of a social-aware fashion recommender system and analyses the impact of user’s personal attributes and profiles on their experiences in various aspects of system use. A user study was conducted to realize the proposed methodology. The obtained insights highlighted that user experiences vary not only from the perspective of using a recommender system but also by varying their personal attributes (age, gender, hobby) and profiles.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
User experience, which encompasses users’ feelings and perceptions, is regarded as a key element in the evaluation of recommender systems. The existing literature extensively works on recommendation generation strategies with focus on the accuracy by considering objective aspects of the system. Although some of the current works considered subjective aspects of the recommendation systems from a user-centric perspective to evaluate the recommender system, however, a comprehensive analysis that could investigate factors to improve user experience was of limited focus. In this paper, we propose a methodology that provides a comprehensive multi-perspective analysis of a social-aware fashion recommender system and analyses the impact of user’s personal attributes and profiles on their experiences in various aspects of system use. A user study was conducted to realize the proposed methodology. The obtained insights highlighted that user experiences vary not only from the perspective of using a recommender system but also by varying their personal attributes (age, gender, hobby) and profiles.
34.
Neidhardt, Julia
Transforming recommender systems: balancing personalization, fairness, and human values Proceedings Article
In: Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, Jeju, Korea, 2024, ISBN: 978-1-956792-04-1.
@inproceedings{10.24963/ijcai.2024/982,
title = {Transforming recommender systems: balancing personalization, fairness, and human values},
author = {Julia Neidhardt},
url = {https://doi.org/10.24963/ijcai.2024/982},
doi = {10.24963/ijcai.2024/982},
isbn = {978-1-956792-04-1},
year = {2024},
date = {2024-01-01},
booktitle = {Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence},
address = {Jeju, Korea},
series = {IJCAI '24},
abstract = {Recent advancements in recommender systems highlight the importance of metrics beyond accuracy, including diversity, serendipity, and fairness. This paper discusses various aspects of modern recommender systems, focusing on challenges such as preference elicitation, the complexity of human decision-making, and multi-domain applicability. The integration of Generative AI and Large Language Models offers enhanced personalization capabilities but also raises concerns regarding transparency and fairness. This work examines ongoing research efforts aimed at developing transparent, fair, and contextually aware systems. Our approach seeks to prioritize user wellbeing and responsibility, contributing to a more equitable and functional digital environment through advanced technologies and interdisciplinary insights.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Recent advancements in recommender systems highlight the importance of metrics beyond accuracy, including diversity, serendipity, and fairness. This paper discusses various aspects of modern recommender systems, focusing on challenges such as preference elicitation, the complexity of human decision-making, and multi-domain applicability. The integration of Generative AI and Large Language Models offers enhanced personalization capabilities but also raises concerns regarding transparency and fairness. This work examines ongoing research efforts aimed at developing transparent, fair, and contextually aware systems. Our approach seeks to prioritize user wellbeing and responsibility, contributing to a more equitable and functional digital environment through advanced technologies and interdisciplinary insights.
35.
Kolb, Thomas Elmar
Enhancing Cross-Domain Recommender Systems with LLMs: Evaluating Bias and Beyond-Accuracy Measures Proceedings Article
In: Proceedings of the 18th ACM Conference on Recommender Systems, pp. 1388–1394, Association for Computing Machinery, Bari, Italy, 2024, ISBN: 9798400705052.
@inproceedings{10.1145/3640457.3688027,
title = {Enhancing Cross-Domain Recommender Systems with LLMs: Evaluating Bias and Beyond-Accuracy Measures},
author = {Thomas Elmar Kolb},
url = {https://doi.org/10.1145/3640457.3688027},
doi = {10.1145/3640457.3688027},
isbn = {9798400705052},
year = {2024},
date = {2024-01-01},
booktitle = {Proceedings of the 18th ACM Conference on Recommender Systems},
pages = {1388–1394},
publisher = {Association for Computing Machinery},
address = {Bari, Italy},
series = {RecSys '24},
abstract = {The research domain of recommender systems is rapidly evolving. Initially, optimization efforts focused primarily on accuracy. However, recent research has highlighted the importance of addressing bias and beyond-accuracy measures such as novelty, diversity, and serendipity. With the rise of multi-domain recommender systems, the need to re-examine bias and beyond-accuracy measures in cross-domain settings has become crucial. Traditional methods face challenges such as cold-start problems, which can potentially be mitigated by leveraging LLMs. This proposed work investigates how LLM-based recommendation methods can enhance cross-domain recommender systems, focusing on identifying, measuring, and mitigating bias while evaluating the impact of beyond-accuracy measures. We aim to provide new insights by comparing traditional and LLM-based systems within a real-world environment encompassing the domains of news, books, and various lifestyle areas. Our research seeks to address the outlined gaps and develop effective evaluation strategies for the unique challenges posed by LLMs in cross-domain recommender systems.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
The research domain of recommender systems is rapidly evolving. Initially, optimization efforts focused primarily on accuracy. However, recent research has highlighted the importance of addressing bias and beyond-accuracy measures such as novelty, diversity, and serendipity. With the rise of multi-domain recommender systems, the need to re-examine bias and beyond-accuracy measures in cross-domain settings has become crucial. Traditional methods face challenges such as cold-start problems, which can potentially be mitigated by leveraging LLMs. This proposed work investigates how LLM-based recommendation methods can enhance cross-domain recommender systems, focusing on identifying, measuring, and mitigating bias while evaluating the impact of beyond-accuracy measures. We aim to provide new insights by comparing traditional and LLM-based systems within a real-world environment encompassing the domains of news, books, and various lifestyle areas. Our research seeks to address the outlined gaps and develop effective evaluation strategies for the unique challenges posed by LLMs in cross-domain recommender systems.
36.
Huebner, Blake; Kolb, Thomas Elmar; Neidhardt, Julia
Evaluating Group Fairness in News Recommendations: A Comparative Study of Algorithms and Metrics Proceedings Article
In: Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization, pp. 337–346, Association for Computing Machinery, Cagliari, Italy, 2024, ISBN: 9798400704666.
@inproceedings{10.1145/3631700.3664897,
title = {Evaluating Group Fairness in News Recommendations: A Comparative Study of Algorithms and Metrics},
author = {Blake Huebner and Thomas Elmar Kolb and Julia Neidhardt},
url = {https://doi.org/10.1145/3631700.3664897},
doi = {10.1145/3631700.3664897},
isbn = {9798400704666},
year = {2024},
date = {2024-01-01},
booktitle = {Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {337–346},
publisher = {Association for Computing Machinery},
address = {Cagliari, Italy},
series = {UMAP Adjunct '24},
abstract = {Beyond accuracy metrics, such as fairness and diversity, have become widely studied topics in recommender systems. Improving these metrics is important not only from an ethical and legal perspective, but can also improve overall user satisfaction. Although these metrics are widely discussed, very little empirical research has been done, especially comparing multiple algorithms across different metrics. This work explores the role of fairness and diversity in news recommender systems, specifically in the context of the Austrian media landscape. This study aims to identify the most effective approaches for generating fair and diverse news recommendations, while addressing the potential negative consequences of biased recommendations and filter bubbles, such as societal polarization and the suppression of information. This includes an extensive literature review of relevant group unfairness metrics and state-of-the-art fairness-aware algorithms. A dataset of articles from an Austrian newspaper was used for empirical research, with analysis performed on fairness, and diversity of recommendations. The key message of the study is that accuracy and fairness can be achieved simultaneously with the right modeling approach, while diversity can be held constant using these modeling techniques. The study recommends the use of Personalized Fairness based on Causal Notion models for accuracy and reducing certain unfairness metrics, and finds Fairness Objectives for Collaborative Filtering models more effective at reducing other types of unfairness. The findings contribute to the field by demonstrating the importance of incorporating these metrics into the design and evaluation of recommender systems.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Beyond accuracy metrics, such as fairness and diversity, have become widely studied topics in recommender systems. Improving these metrics is important not only from an ethical and legal perspective, but can also improve overall user satisfaction. Although these metrics are widely discussed, very little empirical research has been done, especially comparing multiple algorithms across different metrics. This work explores the role of fairness and diversity in news recommender systems, specifically in the context of the Austrian media landscape. This study aims to identify the most effective approaches for generating fair and diverse news recommendations, while addressing the potential negative consequences of biased recommendations and filter bubbles, such as societal polarization and the suppression of information. This includes an extensive literature review of relevant group unfairness metrics and state-of-the-art fairness-aware algorithms. A dataset of articles from an Austrian newspaper was used for empirical research, with analysis performed on fairness, and diversity of recommendations. The key message of the study is that accuracy and fairness can be achieved simultaneously with the right modeling approach, while diversity can be held constant using these modeling techniques. The study recommends the use of Personalized Fairness based on Causal Notion models for accuracy and reducing certain unfairness metrics, and finds Fairness Objectives for Collaborative Filtering models more effective at reducing other types of unfairness. The findings contribute to the field by demonstrating the importance of incorporating these metrics into the design and evaluation of recommender systems.
37.
Nalis, Irina; Sippl, Tobias; Kolb, Thomas Elmar; Neidhardt, Julia
Navigating Serendipity – An Experimental User Study On The Interplay of Trust and Serendipity In Recommender Systems Proceedings Article
In: Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization, pp. 386–393, Association for Computing Machinery, Cagliari, Italy, 2024, ISBN: 9798400704666.
@inproceedings{10.1145/3631700.3664901,
title = {Navigating Serendipity - An Experimental User Study On The Interplay of Trust and Serendipity In Recommender Systems},
author = {Irina Nalis and Tobias Sippl and Thomas Elmar Kolb and Julia Neidhardt},
url = {https://doi.org/10.1145/3631700.3664901},
doi = {10.1145/3631700.3664901},
isbn = {9798400704666},
year = {2024},
date = {2024-01-01},
booktitle = {Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {386–393},
publisher = {Association for Computing Machinery},
address = {Cagliari, Italy},
series = {UMAP Adjunct '24},
abstract = {Recommender systems play a crucial role in our daily lives, constantly evolving to meet the diverse needs of users. As the pursuit of improved user experiences continues, metrics such as serendipity have emerged within the realm of beyond-accuracy paradigms. However, integrating serendipitous recommendations presents complex challenges, necessitating a delicate balance between novelty, relevance, and user engagement. In this interdisciplinary experimental user study, we address these challenges within the context of a book recommender system. By investigating the impact of interface design changes on user trust, a key determinant of satisfaction with serendipitous recommendations, we measured trust levels for both individual recommended items and the recommender system as a whole. Our findings indicate that while interface enhancements did not yield significant increases in trust, they did notably elevate serendipity ratings for previously unknown books. These results highlight the intricate interplay between technical and psychological factors in the design of recommender systems, emphasizing the importance of human-centered approaches in the creation of more responsible AI applications. This research contributes to ongoing discussions surrounding user-centric recommendation systems and aligns with broader themes of digital humanism and responsible AI.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Recommender systems play a crucial role in our daily lives, constantly evolving to meet the diverse needs of users. As the pursuit of improved user experiences continues, metrics such as serendipity have emerged within the realm of beyond-accuracy paradigms. However, integrating serendipitous recommendations presents complex challenges, necessitating a delicate balance between novelty, relevance, and user engagement. In this interdisciplinary experimental user study, we address these challenges within the context of a book recommender system. By investigating the impact of interface design changes on user trust, a key determinant of satisfaction with serendipitous recommendations, we measured trust levels for both individual recommended items and the recommender system as a whole. Our findings indicate that while interface enhancements did not yield significant increases in trust, they did notably elevate serendipity ratings for previously unknown books. These results highlight the intricate interplay between technical and psychological factors in the design of recommender systems, emphasizing the importance of human-centered approaches in the creation of more responsible AI applications. This research contributes to ongoing discussions surrounding user-centric recommendation systems and aligns with broader themes of digital humanism and responsible AI.
38.
Scholz, Felix; Kolb, Thomas Elmar; Neidhardt, Julia
Classifying User Roles in Online News Forums: A Model for User Interaction and Behavior Analysis Proceedings Article
In: Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization, pp. 240–249, Association for Computing Machinery, Cagliari, Italy, 2024, ISBN: 9798400704666.
@inproceedings{10.1145/3631700.3665187,
title = {Classifying User Roles in Online News Forums: A Model for User Interaction and Behavior Analysis},
author = {Felix Scholz and Thomas Elmar Kolb and Julia Neidhardt},
url = {https://doi.org/10.1145/3631700.3665187},
doi = {10.1145/3631700.3665187},
isbn = {9798400704666},
year = {2024},
date = {2024-01-01},
booktitle = {Adjunct Proceedings of the 32nd ACM Conference on User Modeling, Adaptation and Personalization},
pages = {240–249},
publisher = {Association for Computing Machinery},
address = {Cagliari, Italy},
series = {UMAP Adjunct '24},
abstract = {The growing exchange of opinions in online news forums brings together a diverse cross-section of users with varying opinions and motivations. Understanding these behaviors is crucial for unraveling the composition of these large user bases. This study proposes an explainable model aimed at classifying users based on their activity and interaction patterns in online news forums. The model leverages exploratory and statistical data analysis to reveal recurring behaviors and provides a tool to analyze the evolution of large user communities, offering an overview of their composition. The model identifies six active roles: Taciturn, Silent Voter, Regular, Conversationalist, Power User, and Celebrity, and one inactive role, Lurker. The model was evaluated for its predictive power, achieving a macro F1 score of 0.8632, demonstrating its robustness. By applying the model to a long-term dataset from the online news forum derStandard.at, an analysis of role distribution over time was conducted. The results indicated a gradual increase in user activity within the forum. Moreover, the study assessed the co-occurrence of roles in users’ long-term behavior and measured the frequency of role changes. This analysis aimed to determine whether users have consistent roles or exhibit various roles, which may depend on time or context.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
The growing exchange of opinions in online news forums brings together a diverse cross-section of users with varying opinions and motivations. Understanding these behaviors is crucial for unraveling the composition of these large user bases. This study proposes an explainable model aimed at classifying users based on their activity and interaction patterns in online news forums. The model leverages exploratory and statistical data analysis to reveal recurring behaviors and provides a tool to analyze the evolution of large user communities, offering an overview of their composition. The model identifies six active roles: Taciturn, Silent Voter, Regular, Conversationalist, Power User, and Celebrity, and one inactive role, Lurker. The model was evaluated for its predictive power, achieving a macro F1 score of 0.8632, demonstrating its robustness. By applying the model to a long-term dataset from the online news forum derStandard.at, an analysis of role distribution over time was conducted. The results indicated a gradual increase in user activity within the forum. Moreover, the study assessed the co-occurrence of roles in users’ long-term behavior and measured the frequency of role changes. This analysis aimed to determine whether users have consistent roles or exhibit various roles, which may depend on time or context.
39.
Godolja, Dante; Kolb, Thomas Elmar; Neidhardt, Julia
Unlocking the Potential of Content-Based Restaurant Recommender Systems Proceedings Article
In: Berezina, Katerina; Nixon, Lyndon; Tuomi, Aarni (Ed.): Information and Communication Technologies in Tourism 2024, pp. 239–244, Springer Nature Switzerland, Cham, 2024, ISBN: 978-3-031-58839-6.
@inproceedings{10.1007/978-3-031-58839-6_26,
title = {Unlocking the Potential of Content-Based Restaurant Recommender Systems},
author = {Dante Godolja and Thomas Elmar Kolb and Julia Neidhardt},
editor = {Katerina Berezina and Lyndon Nixon and Aarni Tuomi},
isbn = {978-3-031-58839-6},
year = {2024},
date = {2024-01-01},
booktitle = {Information and Communication Technologies in Tourism 2024},
pages = {239–244},
publisher = {Springer Nature Switzerland},
address = {Cham},
abstract = {Content-based restaurant recommender systems use features such as cuisine type, price range, and location to suggest dining options to users. Current research explores ways to improve their effectiveness. In this work, we explore different ideas on how to build a recommender system. We explore TF-IDF as a baseline and the state-of-the-art model SBERT. These ideas are tested on a real-world data-set of a digital restaurant guide. Evaluation involves both qualitative assessment by a domain expert and quantitative analysis. The results show that, with proper preprocessing, TF-IDF can achieve similar scores to SBERT and, depending on the scenario, even better results. However, SBERT still provides more novel recommendations than TF-IDF. Depending on the scenario, both models can be used to generate meaningful restaurant recommendations. However, more implicit aspects like a restaurant's atmosphere can hardly be captured by these models.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Content-based restaurant recommender systems use features such as cuisine type, price range, and location to suggest dining options to users. Current research explores ways to improve their effectiveness. In this work, we explore different ideas on how to build a recommender system. We explore TF-IDF as a baseline and the state-of-the-art model SBERT. These ideas are tested on a real-world data-set of a digital restaurant guide. Evaluation involves both qualitative assessment by a domain expert and quantitative analysis. The results show that, with proper preprocessing, TF-IDF can achieve similar scores to SBERT and, depending on the scenario, even better results. However, SBERT still provides more novel recommendations than TF-IDF. Depending on the scenario, both models can be used to generate meaningful restaurant recommendations. However, more implicit aspects like a restaurant’s atmosphere can hardly be captured by these models.
40.
Nalis-Neuner, Irina
Digital Humanism for Sustainable Solutions to Wicked Problems: A Workshop Exploring Collective Intelligence Design Miscellaneous
2023.
@misc{ 20.500.12708_192558b,
title = {Digital Humanism for Sustainable Solutions to Wicked Problems: A Workshop Exploring Collective Intelligence Design},
author = {Irina Nalis-Neuner},
year = {2023},
date = {2023-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}