Assoc. Prof. Mehdi Elahi
Work Package Leader
University of Bergen
Mehdi Elahi is an Associate Professor at the University of Bergen (UiB), Norway. Before joining this university, in 2014, Mehdi Elahi has obtained his Ph.D. degree in Computer Science and since then, he has published more than 70 peer-reviewed journal and conference publications. His current #citation is 1700+ and his H-index is 20. His research has been mainly focused on AI, Data Science, and Cognitive Science, with an emphasis on their potential industrial applications such as on Recommender Systems. He has also co-invented and co-coowned an AI-related US-patent. Mehdi Elahi has been involved in the authorship of several EU grant proposals such as the large-scale grants, recently funded with a budget of 30 Million Euro, where he will serve as WP Leader for 8 years. Before that, he has received prestigious research credits from giant IT industries (i.e., Amazon and Google).
His research findings have been published in some of the most prestigious reference literature of the field (e.g., Recommender Systems Handbook). One of his journal articles has been the second most cited paper of a top Elsevier journal. He has organized International Data Challenges together with top Companies (i.e., Spotify and XING).
2024
Mahmood, Bilal; Elahi, Mehdi; Touileb, Samia; Steskal, Lubos
Can Large Language Models Support Editors Pick Related News Articles? Conference
NIKT 2024, 2024.
@conference{canlarge24,
title = {Can Large Language Models Support Editors Pick Related News Articles?},
author = {Bilal Mahmood and Mehdi Elahi and Samia Touileb and Lubos Steskal},
url = {https://mediafutures.no/can_large_language_models_support_editors_pick_related_news_articles____nik_2024-1/},
year = {2024},
date = {2024-11-05},
booktitle = {NIKT 2024},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Mahmood, Bilal; Elahi, Mehdi; Vadiee, Fahrhad; Touileb, Samia; Steskal, Lubos
A Supervised Machine Learning Approach for Supporting Editorial Article Selection Working paper
2024.
@workingpaper{supervisedmachin24,
title = {A Supervised Machine Learning Approach for Supporting Editorial Article Selection},
author = {Bilal Mahmood and Mehdi Elahi and Fahrhad Vadiee and Samia Touileb and Lubos Steskal},
url = {https://mediafutures.no/inra_long_paper_for_workshop_2024__ml_/},
year = {2024},
date = {2024-11-05},
issue = {INRA Long paper 2024},
keywords = {},
pubstate = {published},
tppubtype = {workingpaper}
}
Klimashevskaia, Anastasiia; Elahi, Mehdi; Jannach, Dietmar; Trattner, Christoph; Buodd, Simen
Empowering Editors: How Automated Recommendations Support Editorial Article Curation Workshop
RecSys 2024, INRA workshop, 2024.
@workshop{empoweringANA24,
title = {Empowering Editors: How Automated Recommendations Support Editorial Article Curation},
author = {Anastasiia Klimashevskaia and Mehdi Elahi and Dietmar Jannach and Christoph Trattner and Simen Buodd},
url = {https://mediafutures.no/recsys2024-workshops_paper_119-3/},
year = {2024},
date = {2024-10-01},
urldate = {2024-10-01},
booktitle = {RecSys 2024, INRA workshop},
issue = {RecSys 2024, INRA workshop},
abstract = {The application of recommender systems in the news domain has experienced rapid growth in recent years. Various news outlets are proposing a full automation of a newspaper front page through automated recommendation. In this paper, however, we explore the synergy of editorial and algorithmic news curation by analyzing the front page of a real-world news platform, where news articles are either selected automatically by a recommendation algorithm or are selected manually by editors. An investigation of the interaction log data from an online newspaper revealed that while the editorial staff is focusing on content that is generally popular across large parts of the audience, the algorithmic curation can, in addition, provide small yet noteworthy personalization touches for individual readers. The results of the analysis demonstrate an example of a successful coexistence of editorial and algorithmic news curation.},
howpublished = {RecSys 2024, INRA workshop},
keywords = {},
pubstate = {published},
tppubtype = {workshop}
}
The application of recommender systems in the news domain has experienced rapid growth in recent years. Various news outlets are proposing a full automation of a newspaper front page through automated recommendation. In this paper, however, we explore the synergy of editorial and algorithmic news curation by analyzing the front page of a real-world news platform, where news articles are either selected automatically by a recommendation algorithm or are selected manually by editors. An investigation of the interaction log data from an online newspaper revealed that while the editorial staff is focusing on content that is generally popular across large parts of the audience, the algorithmic curation can, in addition, provide small yet noteworthy personalization touches for individual readers. The results of the analysis demonstrate an example of a successful coexistence of editorial and algorithmic news curation.
Majjodi, Ayoub El; Khan, Sohail Ahmed; Starke, Alain D.; Elahi, Mehdi; Trattner, Christoph
Advancing Visual Food Attractiveness Predictions for Healthy Food Recommender Systems Conference
The ACM Conference on Recommender Systems (RecSys) 2024, 2024.
@conference{visualfood24,
title = {Advancing Visual Food Attractiveness Predictions for Healthy Food Recommender Systems},
author = {Ayoub El Majjodi and Sohail Ahmed Khan and Alain D. Starke and Mehdi Elahi and Christoph Trattner},
url = {https://mediafutures.no/healthrecsys-2024-ayoub-1/},
year = {2024},
date = {2024-09-17},
booktitle = {The ACM Conference on Recommender Systems (RecSys) 2024},
abstract = {The visual representation of food has a significant influence on
how people choose food in the real world but also in a digital food
recommender scenario. Previous studies on that matter show that
small change in visual features can change human decision-making,
regardless of whether the food is healthy or not. This paper reports
on a study that aims to understand further how users perceive
the attractiveness of food images in the digital world. In an online
mixed-methods survey (N=192), users provided visual attractive-
ness ratings on a 7-point scale and provided textual assessments
of the visual attractiveness of food images. We found a robust
correlation between fundamental visual features (e.g., contrast, col-
orfulness) and perceived image attractiveness. The analysis also
revealed that cooking skills predicted food image attractiveness
among user factors. Regarding food image dimensions, appearance
and perceived healthiness emerged to be significantly correlated
with user ratings for food image attractiveness.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
The visual representation of food has a significant influence on
how people choose food in the real world but also in a digital food
recommender scenario. Previous studies on that matter show that
small change in visual features can change human decision-making,
regardless of whether the food is healthy or not. This paper reports
on a study that aims to understand further how users perceive
the attractiveness of food images in the digital world. In an online
mixed-methods survey (N=192), users provided visual attractive-
ness ratings on a 7-point scale and provided textual assessments
of the visual attractiveness of food images. We found a robust
correlation between fundamental visual features (e.g., contrast, col-
orfulness) and perceived image attractiveness. The analysis also
revealed that cooking skills predicted food image attractiveness
among user factors. Regarding food image dimensions, appearance
and perceived healthiness emerged to be significantly correlated
with user ratings for food image attractiveness.
how people choose food in the real world but also in a digital food
recommender scenario. Previous studies on that matter show that
small change in visual features can change human decision-making,
regardless of whether the food is healthy or not. This paper reports
on a study that aims to understand further how users perceive
the attractiveness of food images in the digital world. In an online
mixed-methods survey (N=192), users provided visual attractive-
ness ratings on a 7-point scale and provided textual assessments
of the visual attractiveness of food images. We found a robust
correlation between fundamental visual features (e.g., contrast, col-
orfulness) and perceived image attractiveness. The analysis also
revealed that cooking skills predicted food image attractiveness
among user factors. Regarding food image dimensions, appearance
and perceived healthiness emerged to be significantly correlated
with user ratings for food image attractiveness.
Mahmood, Bilal; Elahi, Mehdi; Touileb, Samia; Steskal, Lubos; Trattner, Christoph
Incorporating Editorial Feedback in the Evaluation of News Recommender Systems Conference
ACM UMAP 2024, 2024.
@conference{incoed24,
title = {Incorporating Editorial Feedback in the Evaluation of News Recommender Systems},
author = {Bilal Mahmood and Mehdi Elahi and Samia Touileb and Lubos Steskal and Christoph Trattner},
url = {https://mediafutures.no/lbr_umap_editorial_component_in_nrs/},
year = {2024},
date = {2024-07-01},
urldate = {2024-07-01},
booktitle = {ACM UMAP 2024},
abstract = {Research in the recommender systems field typically applies a rather traditional evaluation methodology when assessing the quality of recommendations. This methodology heavily relies on incorporating different forms of user feedback (e.g., clicks) representing the specific needs and interests of the users. While this methodology may offer various benefits, it may fail to comprehensively project the complexities of certain application domains, such as the news domain. This domain is distinct from other domains primarily due to the strong influence of editorial control in the news delivery process. Incorporation of this role can profoundly impact how the relevance of news articles is measured when recommended to the users. Despite its critical importance, there appears to be a research gap in investigating the dynamics between the roles of editorial control and personalization in the community of recommender systems. In this paper, we address this gap by conducting experiments where the relevance of recommendations is assessed from an editorial perspective. We received a real-world dataset from TV 2, one of the largest editor-managed commercial media houses in Norway, which includes editors’ feedback on how news articles are being related. In our experiment, we considered a scenario where algorithm-generated recommendations, using the K-Nearest Neighbor (KNN) model, employing various text embedding models to encode different sections of the news articles (e.g., title, lead title, body text, and full text), are compared against the editorial feedback. The results are promising, demonstrating the effectiveness of the recommendation in fulfilling the editorial prospects.},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Research in the recommender systems field typically applies a rather traditional evaluation methodology when assessing the quality of recommendations. This methodology heavily relies on incorporating different forms of user feedback (e.g., clicks) representing the specific needs and interests of the users. While this methodology may offer various benefits, it may fail to comprehensively project the complexities of certain application domains, such as the news domain. This domain is distinct from other domains primarily due to the strong influence of editorial control in the news delivery process. Incorporation of this role can profoundly impact how the relevance of news articles is measured when recommended to the users. Despite its critical importance, there appears to be a research gap in investigating the dynamics between the roles of editorial control and personalization in the community of recommender systems. In this paper, we address this gap by conducting experiments where the relevance of recommendations is assessed from an editorial perspective. We received a real-world dataset from TV 2, one of the largest editor-managed commercial media houses in Norway, which includes editors’ feedback on how news articles are being related. In our experiment, we considered a scenario where algorithm-generated recommendations, using the K-Nearest Neighbor (KNN) model, employing various text embedding models to encode different sections of the news articles (e.g., title, lead title, body text, and full text), are compared against the editorial feedback. The results are promising, demonstrating the effectiveness of the recommendation in fulfilling the editorial prospects.
Klimashevskaia, Anastasiia; Jannach, Dietmar; Elahi, Mehdi; Trattner, Christoph
A Survey on Popularity Bias in Recommender Systems Journal Article Forthcoming
In: User Modeling and User-Adapted Interaction (UMUAI), Forthcoming.
@article{biasanas24,
title = {A Survey on Popularity Bias in Recommender Systems},
author = {Anastasiia Klimashevskaia and Dietmar Jannach and Mehdi Elahi and Christoph Trattner},
url = {https://mediafutures.no/popularitybias_literature_review-5/},
year = {2024},
date = {2024-06-13},
journal = {User Modeling and User-Adapted Interaction (UMUAI)},
abstract = {Recommender systems help people find relevant content in a personalized way. One main promise of such systems is that they are able to increase the visibility of items in the long tail, i.e., the lesser-
known items in a catalogue. Existing research, however, suggests that in many situations today’s recommendation algorithms instead exhibit a popularity bias, meaning that they often focus on rather popular items in their recommendations. Such a bias may not only lead to the limited value of the recommendations for consumers and providers in the short run, but it may also cause undesired reinforcement effects over time. In this paper, we discuss the potential reasons for popularity bias and review existing approaches to detect, quantify and mitigate popularity bias in
recommender systems. Our survey, therefore, includes both an overview
of the computational metrics used in the literature as well as a review of the main technical approaches to reduce the bias. Furthermore, we critically discuss today’s literature, where we observe that the research is almost entirely based on computational experiments and on certain
assumptions regarding the practical effects of including long-tail items in the recommendations.},
keywords = {},
pubstate = {forthcoming},
tppubtype = {article}
}
Recommender systems help people find relevant content in a personalized way. One main promise of such systems is that they are able to increase the visibility of items in the long tail, i.e., the lesser-
known items in a catalogue. Existing research, however, suggests that in many situations today’s recommendation algorithms instead exhibit a popularity bias, meaning that they often focus on rather popular items in their recommendations. Such a bias may not only lead to the limited value of the recommendations for consumers and providers in the short run, but it may also cause undesired reinforcement effects over time. In this paper, we discuss the potential reasons for popularity bias and review existing approaches to detect, quantify and mitigate popularity bias in
recommender systems. Our survey, therefore, includes both an overview
of the computational metrics used in the literature as well as a review of the main technical approaches to reduce the bias. Furthermore, we critically discuss today’s literature, where we observe that the research is almost entirely based on computational experiments and on certain
assumptions regarding the practical effects of including long-tail items in the recommendations.
known items in a catalogue. Existing research, however, suggests that in many situations today’s recommendation algorithms instead exhibit a popularity bias, meaning that they often focus on rather popular items in their recommendations. Such a bias may not only lead to the limited value of the recommendations for consumers and providers in the short run, but it may also cause undesired reinforcement effects over time. In this paper, we discuss the potential reasons for popularity bias and review existing approaches to detect, quantify and mitigate popularity bias in
recommender systems. Our survey, therefore, includes both an overview
of the computational metrics used in the literature as well as a review of the main technical approaches to reduce the bias. Furthermore, we critically discuss today’s literature, where we observe that the research is almost entirely based on computational experiments and on certain
assumptions regarding the practical effects of including long-tail items in the recommendations.
2023
Klimashevskaia, Anastasiia; Elahi, Mehdi; Jannach, Dietmar; Skjærven, Lars; Tessem, Astrid; Trattner, Christoph
Evaluating The Effects of Calibrated Popularity Bias Mitigation: A Field Study Conference
Association for Computing Machinery (ACM) RecSys ’23, 2023.
@conference{RecSys_2023_LBR,
title = {Evaluating The Effects of Calibrated Popularity Bias Mitigation: A Field Study},
author = {Anastasiia Klimashevskaia and Mehdi Elahi and Dietmar Jannach and Lars Skjærven and Astrid Tessem and Christoph Trattner},
url = {https://mediafutures.no/recsys_2023_lbr/},
year = {2023},
date = {2023-09-18},
booktitle = {Association for Computing Machinery (ACM) RecSys ’23},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
2022
Elahi, Mehdi; Starke, Alain D.; Ioini, Nabil El; Lambrix, Anna Alexander; Trattner, Christoph
Developing and Evaluating a University Recommender System Journal Article
In: Frontiers in Artificial Intelligence , 2022.
@article{Elahi2022,
title = {Developing and Evaluating a University Recommender System},
author = {Mehdi Elahi and Alain D. Starke and Nabil El Ioini and Anna Alexander Lambrix and Christoph Trattner},
url = {https://www.frontiersin.org/articles/10.3389/frai.2021.796268/full},
doi = {https://doi.org/10.3389/frai.2021.796268},
year = {2022},
date = {2022-02-02},
journal = {Frontiers in Artificial Intelligence },
abstract = {A challenge for many young adults is to find the right institution to follow higher education. Global university rankings are a commonly used, but inefficient tool, for they do not consider a person's preferences and needs. For example, some persons pursue prestige in their higher education, while others prefer proximity. This paper develops and evaluates a university recommender system, eliciting user preferences as ratings to build predictive models and to generate personalized university ranking lists. In Study 1, we performed offline evaluation on a rating dataset to determine which recommender approaches had the highest predictive value. In Study 2, we selected three algorithms to produce different university recommendation lists in our online tool, asking our users to compare and evaluate them in terms of different metrics (Accuracy, Diversity, Perceived Personalization, Satisfaction, and Novelty). We show that a SVD algorithm scores high on accuracy and perceived personalization, while a KNN algorithm scores better on novelty. We also report findings on preferred university features.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
A challenge for many young adults is to find the right institution to follow higher education. Global university rankings are a commonly used, but inefficient tool, for they do not consider a person's preferences and needs. For example, some persons pursue prestige in their higher education, while others prefer proximity. This paper develops and evaluates a university recommender system, eliciting user preferences as ratings to build predictive models and to generate personalized university ranking lists. In Study 1, we performed offline evaluation on a rating dataset to determine which recommender approaches had the highest predictive value. In Study 2, we selected three algorithms to produce different university recommendation lists in our online tool, asking our users to compare and evaluate them in terms of different metrics (Accuracy, Diversity, Perceived Personalization, Satisfaction, and Novelty). We show that a SVD algorithm scores high on accuracy and perceived personalization, while a KNN algorithm scores better on novelty. We also report findings on preferred university features.
Kvifte, Tord; Elahi, Mehdi; Trattner, Christoph
Hybrid Recommendation of Movies based on Deep Content Features Proceedings Article
In: Springer Nature, 2022.
@inproceedings{cristin1957037,
title = {Hybrid Recommendation of Movies based on Deep Content Features},
author = {Tord Kvifte and Mehdi Elahi and Christoph Trattner},
url = {https://app.cristin.no/results/show.jsf?id=1957037, Cristin
https://aip-research-center.github.io/AIPA_workshop/2021/},
year = {2022},
date = {2022-01-01},
booktitle = {Springer Nature},
abstract = {When a movie is uploaded to a movie Recommender System (e.g., YouTube), the system can exploit various forms of descriptive features (e.g., tags and genre) in order to generate personalized recommendation for users. However, there are situations where the descriptive features are missing or very limited and the system may fail to include such a movie in the recommendation list. This paper investigates hybrid recommendation based on a novel form of content features, extracted from movies, in order to generate recommendation for users. Such features represent the visual aspects of movies, based on Deep Learning models, and hence, do not require any human annotation when extracted. We have evaluated our proposed technique using a large dataset of movies and shown that automatically extracted visual features can mitigate the cold-start problem by generating recommendation with a superior quality compared to different baselines, including recommendation based on human-annotated features.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
When a movie is uploaded to a movie Recommender System (e.g., YouTube), the system can exploit various forms of descriptive features (e.g., tags and genre) in order to generate personalized recommendation for users. However, there are situations where the descriptive features are missing or very limited and the system may fail to include such a movie in the recommendation list. This paper investigates hybrid recommendation based on a novel form of content features, extracted from movies, in order to generate recommendation for users. Such features represent the visual aspects of movies, based on Deep Learning models, and hence, do not require any human annotation when extracted. We have evaluated our proposed technique using a large dataset of movies and shown that automatically extracted visual features can mitigate the cold-start problem by generating recommendation with a superior quality compared to different baselines, including recommendation based on human-annotated features.
2021
Trattner, Christoph; Jannach, Dietmar; Motta, Enrico; Meijer, Irene Costera; Diakopoulos, Nicholas; Elahi, Mehdi; Opdahl, Andreas L.; Tessem, Bjørnar; Borch, Njål; Fjeld, Morten; Øvrelid, Lilja; Smedt, Koenraad De; Moe, Hallvard
Responsible media technology and AI: challenges and research directions Journal Article
In: AI and Ethics, 2021.
@article{cristin2000622,
title = {Responsible media technology and AI: challenges and research directions},
author = {Christoph Trattner and Dietmar Jannach and Enrico Motta and Irene Costera Meijer and Nicholas Diakopoulos and Mehdi Elahi and Andreas L. Opdahl and Bjørnar Tessem and Njål Borch and Morten Fjeld and Lilja Øvrelid and Koenraad De Smedt and Hallvard Moe},
url = {https://app.cristin.no/results/show.jsf?id=2000622, Cristin
https://link.springer.com/content/pdf/10.1007/s43681-021-00126-4.pdf},
doi = {https://doi.org/10.1007/s43681-021-00126-4},
year = {2021},
date = {2021-12-20},
urldate = {2021-12-20},
journal = {AI and Ethics},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Elahi, Mehdi; Jannach, Dietmar; Skjærven, Lars; Knudsen, Erik; Sjøvaag, Helle; Tolonen, Kristian; Holmstad, Øyvind; Pipkin, Igor; Throndsen, Eivind; Stenbom, Agnes; Fiskerud, Eivind; Oesch, Adrian; Vredenberg, Loek; Trattner, Christoph
Towards Responsible Media Recommendation Journal Article
In: AI and Ethics, 2021.
@article{cristin1953352,
title = {Towards Responsible Media Recommendation},
author = {Mehdi Elahi and Dietmar Jannach and Lars Skjærven and Erik Knudsen and Helle Sjøvaag and Kristian Tolonen and Øyvind Holmstad and Igor Pipkin and Eivind Throndsen and Agnes Stenbom and Eivind Fiskerud and Adrian Oesch and Loek Vredenberg and Christoph Trattner},
url = {https://app.cristin.no/results/show.jsf?id=1953352, Cristin
https://link.springer.com/article/10.1007%2Fs43681-021-00107-7},
doi = {https://doi.org/10.1007/s43681-021-00107-7},
year = {2021},
date = {2021-11-02},
journal = {AI and Ethics},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Elahi, Mehdi; Kvifte, Tord
Video Recommendations Based on Visual Features Extracted with Deep Learning Masters Thesis
Universitetet i Bergen, 2021.
@mastersthesis{cristin1956990,
title = {Video Recommendations Based on Visual Features Extracted with Deep Learning},
author = {Mehdi Elahi and Tord Kvifte},
url = {https://app.cristin.no/results/show.jsf?id=1956990, Cristin
https://hdl.handle.net/11250/2760300},
year = {2021},
date = {2021-01-01},
school = {Universitetet i Bergen},
abstract = {When a movie is uploaded to a movie Recommender System (e.g., YouTube), the system can exploit various forms of descriptive features (e.g., tags and genre) in order to generate personalized recommendation for users. However, there are situations where the descriptive features are missing or very limited and the system may fail to include such a movie in the recommendation list, known as Cold-start problem. This thesis investigates recommendation based on a novel form of content features, extracted from movies, in order to generate recommendation for users. Such features represent the visual aspects of movies, based on Deep Learning models, and hence, do not require any human annotation when extracted. The proposed technique has been evaluated in both offline and online evaluations using a large dataset of movies. The online evaluation has been carried out in a evaluation framework developed for this thesis. Results from the offline and online evaluation (N=150) show that automatically extracted visual features can mitigate the cold-start problem by generating recommendation with a superior quality compared to different baselines, including recommendation based on human-annotated features. The results also point to subtitles as a high-quality future source of automatically extracted features. The visual feature dataset, named DeepCineProp13K and the subtitle dataset, CineSub3K, as well as the proposed evaluation framework are all made openly available online in a designated Github repository.},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
When a movie is uploaded to a movie Recommender System (e.g., YouTube), the system can exploit various forms of descriptive features (e.g., tags and genre) in order to generate personalized recommendation for users. However, there are situations where the descriptive features are missing or very limited and the system may fail to include such a movie in the recommendation list, known as Cold-start problem. This thesis investigates recommendation based on a novel form of content features, extracted from movies, in order to generate recommendation for users. Such features represent the visual aspects of movies, based on Deep Learning models, and hence, do not require any human annotation when extracted. The proposed technique has been evaluated in both offline and online evaluations using a large dataset of movies. The online evaluation has been carried out in a evaluation framework developed for this thesis. Results from the offline and online evaluation (N=150) show that automatically extracted visual features can mitigate the cold-start problem by generating recommendation with a superior quality compared to different baselines, including recommendation based on human-annotated features. The results also point to subtitles as a high-quality future source of automatically extracted features. The visual feature dataset, named DeepCineProp13K and the subtitle dataset, CineSub3K, as well as the proposed evaluation framework are all made openly available online in a designated Github repository.
Elahi, Mehdi; Moghaddam, Farshad Bakhshandegan; Hosseini, Reza; Rimaz, Mohammad Hossein; Trattner, Christoph
Enhanced Movie Recommendation Incorporating Visual Features Working paper
2021.
@workingpaper{cristin1957034,
title = {Enhanced Movie Recommendation Incorporating Visual Features},
author = {Mehdi Elahi and Farshad Bakhshandegan Moghaddam and Reza Hosseini and Mohammad Hossein Rimaz and Christoph Trattner},
url = {https://app.cristin.no/results/show.jsf?id=1957034, Cristin},
year = {2021},
date = {2021-01-01},
keywords = {},
pubstate = {published},
tppubtype = {workingpaper}
}
Elahi, Mehdi; Abdollahpouri, Himan; Mansoury, Masoud; Torkamaan, Helma
Beyond Algorithmic Fairness in Recommender System Working paper
2021.
@workingpaper{cristin1957035,
title = {Beyond Algorithmic Fairness in Recommender System},
author = {Mehdi Elahi and Himan Abdollahpouri and Masoud Mansoury and Helma Torkamaan},
url = {https://app.cristin.no/results/show.jsf?id=1957035, Cristin
https://dl.acm.org/doi/abs/10.1145/3450614.3461685},
year = {2021},
date = {2021-01-01},
keywords = {},
pubstate = {published},
tppubtype = {workingpaper}
}
Elahi, Mehdi; Johannessen, Øyvind
Novel Methods Using Human Emotion and Visual Features for Recommending Movies Masters Thesis
Universitetet i Bergen, 2021.
@mastersthesis{cristin1957008,
title = {Novel Methods Using Human Emotion and Visual Features for Recommending Movies},
author = {Mehdi Elahi and Øyvind Johannessen},
url = {https://app.cristin.no/results/show.jsf?id=1957008, Cristin},
year = {2021},
date = {2021-01-01},
school = {Universitetet i Bergen},
abstract = {This master thesis investigates novel methods using human emotion as contextual information to estimate and elicit ratings when watching movie trailers. The aim is to acquire user preferences without the intrusive and time-consuming behavior of Explicit Feedback strategies, and generate quality recommendations. The proposed preference-elicitation technique is implemented as an Emotion-based Filtering technique (EF) to generate recommendations, and is evaluated against two other recommendation techniques. One Visual-based Filtering technique, using low-level visual features of movies, and one Collaborative Filtering (CF) using explicit ratings. In terms of Accuracy, we found the Emotion-based Filtering technique (EF) to perform better than the two other filtering techniques. In terms of Diversity, the Visual-based Filtering (VF) performed best. We further analyse the obtained data to see if movie genres tend to induce specific emotions, and the potential correlation between emotional responses of users and visual features of movie trailers. When investigating emotional responses, we found that joy and disgust tend to be more prominent in movie genres than other emotions. Our findings also suggest potential correlations on a per movie level. The proposed Emotion-based Filtering technique can be adopted as an Implicit Feedback strategy to obtain user preferences. For future work, we will extend the experiment with more participants and build stronger affective profiles to be studied when recommending movies.},
keywords = {},
pubstate = {published},
tppubtype = {mastersthesis}
}
This master thesis investigates novel methods using human emotion as contextual information to estimate and elicit ratings when watching movie trailers. The aim is to acquire user preferences without the intrusive and time-consuming behavior of Explicit Feedback strategies, and generate quality recommendations. The proposed preference-elicitation technique is implemented as an Emotion-based Filtering technique (EF) to generate recommendations, and is evaluated against two other recommendation techniques. One Visual-based Filtering technique, using low-level visual features of movies, and one Collaborative Filtering (CF) using explicit ratings. In terms of Accuracy, we found the Emotion-based Filtering technique (EF) to perform better than the two other filtering techniques. In terms of Diversity, the Visual-based Filtering (VF) performed best. We further analyse the obtained data to see if movie genres tend to induce specific emotions, and the potential correlation between emotional responses of users and visual features of movie trailers. When investigating emotional responses, we found that joy and disgust tend to be more prominent in movie genres than other emotions. Our findings also suggest potential correlations on a per movie level. The proposed Emotion-based Filtering technique can be adopted as an Implicit Feedback strategy to obtain user preferences. For future work, we will extend the experiment with more participants and build stronger affective profiles to be studied when recommending movies.
Abdollahpouri, Himan; Elahi, Mehdi; Mansoury, Masoud; Sahebi, Shaghayegh; Nazari, Zahra; Chaney, Allison; Loni, Babak
MORS 2021: 1st Workshop on Multi Objective Recommender Systems Proceedings Article
In: Association for Computing Machinery (ACM), 2021.
@inproceedings{cristin1956978,
title = {MORS 2021: 1st Workshop on Multi Objective Recommender Systems},
author = {Himan Abdollahpouri and Mehdi Elahi and Masoud Mansoury and Shaghayegh Sahebi and Zahra Nazari and Allison Chaney and Babak Loni},
url = {https://app.cristin.no/results/show.jsf?id=1956978, Cristin
https://dl.acm.org/doi/10.1145/3460231.3470936},
year = {2021},
date = {2021-01-01},
booktitle = {Association for Computing Machinery (ACM)},
abstract = {Historically, the main criterion for a successful recommender system was the relevance of the recommended items to the user. In other words, the only objective for the recommendation algorithm was to learn user’s preferences for different items and generate recommendations accordingly. However, real-world recommender systems are well beyond a simple objective and often need to take into account multiple objectives simultaneously. These objectives can be either from the users’ perspective or they could come from other stakeholders such as item providers or any party that could be impacted by the recommendations. Such multi-objective and multi-stakeholder recommenders present unique challenges and these challenges were the focus of the MORS workshop.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Historically, the main criterion for a successful recommender system was the relevance of the recommended items to the user. In other words, the only objective for the recommendation algorithm was to learn user’s preferences for different items and generate recommendations accordingly. However, real-world recommender systems are well beyond a simple objective and often need to take into account multiple objectives simultaneously. These objectives can be either from the users’ perspective or they could come from other stakeholders such as item providers or any party that could be impacted by the recommendations. Such multi-objective and multi-stakeholder recommenders present unique challenges and these challenges were the focus of the MORS workshop.
Elahi, Mehdi; Kholgh, Danial Khosh; Kiarostami, Mohammad Sina; Saghari, Sorush; Rad, Shiva Parsa; Tkalcic, Marko
Investigating the impact of recommender systems on user-based and item-based popularity bias Journal Article
In: Information Processing & Management, 2021.
@article{cristin1953363,
title = {Investigating the impact of recommender systems on user-based and item-based popularity bias},
author = {Mehdi Elahi and Danial Khosh Kholgh and Mohammad Sina Kiarostami and Sorush Saghari and Shiva Parsa Rad and Marko Tkalcic},
url = {https://app.cristin.no/results/show.jsf?id=1953363, Cristin
https://www.sciencedirect.com/science/article/pii/S0306457321001436},
doi = {https://doi.org/10.1016/j.ipm.2021.102655},
year = {2021},
date = {2021-01-01},
journal = {Information Processing & Management},
abstract = {Recommender Systems are decision support tools that adopt advanced algorithms in order to help users to find less-explored items that can be interesting for them. While recommender systems may offer a range of attractive benefits, they may also intensify undesired effects, such as the Popularity Bias, where a few popular users/items get more popular and many unpopular users/items get more unpopular.
In this paper, we study the impact of different recommender algorithms on the popularity bias in different application domains and recommendation scenarios. We have designed a comprehensive evaluation methodology by considering two different recommendation scenarios, i.e., the user-based scenario (e.g., recommending users to users to follow), and the item-based scenario (e.g., recommending items to users to consume). We have used two large datasets, Twitter and Movielens, and compared a wide range of classical and modern recommender algorithms by considering a diverse range of metrics, such as PR-AUC, RCE, Gini index, and Entropy Score.
The results have shown a substantial difference between different scenarios and different recommendation domains. According to our observations, while the recommendation of users to users may increase the popularity bias in the system, the recommendation of items to users may indeed decrease it. Moreover, while we have measured a different level of popularity bias in different languages (i.e., English, Spanish, Portuguese, and Japaneses), the above-noted phenomena has been consistently observed in all of these languages.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Recommender Systems are decision support tools that adopt advanced algorithms in order to help users to find less-explored items that can be interesting for them. While recommender systems may offer a range of attractive benefits, they may also intensify undesired effects, such as the Popularity Bias, where a few popular users/items get more popular and many unpopular users/items get more unpopular.
In this paper, we study the impact of different recommender algorithms on the popularity bias in different application domains and recommendation scenarios. We have designed a comprehensive evaluation methodology by considering two different recommendation scenarios, i.e., the user-based scenario (e.g., recommending users to users to follow), and the item-based scenario (e.g., recommending items to users to consume). We have used two large datasets, Twitter and Movielens, and compared a wide range of classical and modern recommender algorithms by considering a diverse range of metrics, such as PR-AUC, RCE, Gini index, and Entropy Score.
The results have shown a substantial difference between different scenarios and different recommendation domains. According to our observations, while the recommendation of users to users may increase the popularity bias in the system, the recommendation of items to users may indeed decrease it. Moreover, while we have measured a different level of popularity bias in different languages (i.e., English, Spanish, Portuguese, and Japaneses), the above-noted phenomena has been consistently observed in all of these languages.
In this paper, we study the impact of different recommender algorithms on the popularity bias in different application domains and recommendation scenarios. We have designed a comprehensive evaluation methodology by considering two different recommendation scenarios, i.e., the user-based scenario (e.g., recommending users to users to follow), and the item-based scenario (e.g., recommending items to users to consume). We have used two large datasets, Twitter and Movielens, and compared a wide range of classical and modern recommender algorithms by considering a diverse range of metrics, such as PR-AUC, RCE, Gini index, and Entropy Score.
The results have shown a substantial difference between different scenarios and different recommendation domains. According to our observations, while the recommendation of users to users may increase the popularity bias in the system, the recommendation of items to users may indeed decrease it. Moreover, while we have measured a different level of popularity bias in different languages (i.e., English, Spanish, Portuguese, and Japaneses), the above-noted phenomena has been consistently observed in all of these languages.
Elahi, Mehdi; Moghaddam, Farshad Bakhshandegan; Hosseini, Reza; Rimaz, Mohammad Hossein; Ioini, Nabil El; Tkalcic, Marko; Trattner, Christoph; Tillo, Tammam
Recommending Videos in Cold Start With Automatic Visual Tags Proceedings Article
In: Association for Computing Machinery (ACM), 2021.
@inproceedings{cristin1956967,
title = {Recommending Videos in Cold Start With Automatic Visual Tags},
author = {Mehdi Elahi and Farshad Bakhshandegan Moghaddam and Reza Hosseini and Mohammad Hossein Rimaz and Nabil El Ioini and Marko Tkalcic and Christoph Trattner and Tammam Tillo},
url = {https://app.cristin.no/results/show.jsf?id=1956967, Cristin
https://dl.acm.org/doi/10.1145/3450614.3461687},
year = {2021},
date = {2021-01-01},
booktitle = {Association for Computing Machinery (ACM)},
abstract = {This paper addresses the so-called New Item problem in video Recommender Systems, as part of Cold Start. New item problem occurs when a new item is added to the system catalog, and the recommender system has no or little data describing that item. This could cause the system to fail to meaningfully recommend the new item to the users. We propose a novel technique that can generate cold start recommendation by utilizing automatic visual tags, i.e., tags that are automatically annotated by deeply analyzing the content of the videos and detecting faces, objects, and even celebrities within the videos. The automatic visual tags do not need any human involvement and have been shown to be very effective in representing the video content. In order to evaluate our proposed technique, we have performed a set of experiments using a large dataset of videos. The results have shown that the automatically extracted visual tags can be incorporated into the cold start recommendation process and achieve superior results compared to the recommendation based on human-annotated tags.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper addresses the so-called New Item problem in video Recommender Systems, as part of Cold Start. New item problem occurs when a new item is added to the system catalog, and the recommender system has no or little data describing that item. This could cause the system to fail to meaningfully recommend the new item to the users. We propose a novel technique that can generate cold start recommendation by utilizing automatic visual tags, i.e., tags that are automatically annotated by deeply analyzing the content of the videos and detecting faces, objects, and even celebrities within the videos. The automatic visual tags do not need any human involvement and have been shown to be very effective in representing the video content. In order to evaluate our proposed technique, we have performed a set of experiments using a large dataset of videos. The results have shown that the automatically extracted visual tags can be incorporated into the cold start recommendation process and achieve superior results compared to the recommendation based on human-annotated tags.
2020
Hazrati, Naieme; Elahi, Mehdi
Addressing the New Item problem in video recommender systems by incorporation of visual features with restricted Boltzmann machines. Journal Article
In: Expert Systems, vol. e12645, pp. 1-20, 2020, (Pre SFI).
@article{Hazrati2020,
title = { Addressing the New Item problem in video recommender systems by incorporation of visual features with restricted Boltzmann machines.},
author = {Naieme Hazrati and Mehdi Elahi},
url = {https://onlinelibrary.wiley.com/doi/epdf/10.1111/exsy.12645},
doi = {https://doi.org/10.1111/exsy.12645},
year = {2020},
date = {2020-10-19},
journal = {Expert Systems},
volume = {e12645},
pages = {1-20},
abstract = {Over the past years, the research of video recommender systems (RSs) has been mainly focussed on the development of novel algorithms. Although beneficial, still any algorithm may fail to recommend video items that the system has no form of data associated to them (New Item Cold Start). This problem occurs when a new item is added to the catalogue of the system and no data are available for that item. In content‐based RSs, the video items are typically represented by semantic attributes, when generating recommendations. These attributes require a group of experts or users for annotation, and still, the generated recommendations might not capture a complete picture of the users' preferences, for example, the visual tastes of users on video style. This article addresses this problem by proposing recommendation based on novel visual features that do not require human annotation and can represent visual aspects of video items. We have designed a novel evaluation methodology considering three realistic scenarios, that is, (a) extreme cold start, (b) moderate cold start and (c) warm‐start scenario. We have conducted a set of comprehensive experiments, and our results have shown the superior performance of recommendations based on visual features, in all of the evaluation scenarios.},
note = {Pre SFI},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Over the past years, the research of video recommender systems (RSs) has been mainly focussed on the development of novel algorithms. Although beneficial, still any algorithm may fail to recommend video items that the system has no form of data associated to them (New Item Cold Start). This problem occurs when a new item is added to the catalogue of the system and no data are available for that item. In content‐based RSs, the video items are typically represented by semantic attributes, when generating recommendations. These attributes require a group of experts or users for annotation, and still, the generated recommendations might not capture a complete picture of the users' preferences, for example, the visual tastes of users on video style. This article addresses this problem by proposing recommendation based on novel visual features that do not require human annotation and can represent visual aspects of video items. We have designed a novel evaluation methodology considering three realistic scenarios, that is, (a) extreme cold start, (b) moderate cold start and (c) warm‐start scenario. We have conducted a set of comprehensive experiments, and our results have shown the superior performance of recommendations based on visual features, in all of the evaluation scenarios.
2017
Elahi, Mehdi; Deldjoo, Yashar; Moghaddam, Farshad Bakhshandegan; Cella, Leonardo; Cerada, Stefano; Cremonesi, Paolo
Exploring the semantic gap for movie recommendations Conference
Proceedings of the Eleventh ACM Conference on Recommender Systems, Association for Computing Machinery New York, 2017, (Pre SFI).
@conference{Elahi2017,
title = {Exploring the semantic gap for movie recommendations},
author = {Mehdi Elahi and Yashar Deldjoo and Farshad Bakhshandegan Moghaddam and Leonardo Cella and Stefano Cerada and Paolo Cremonesi },
url = {https://dl.acm.org/doi/pdf/10.1145/3109859.3109908},
doi = {https://doi.org/10.1145/3109859.3109908},
year = {2017},
date = {2017-08-01},
booktitle = {Proceedings of the Eleventh ACM Conference on Recommender Systems},
pages = {326–330},
address = {New York},
organization = {Association for Computing Machinery},
abstract = {In the last years, there has been much attention given to the semantic gap problem in multimedia retrieval systems. Much effort has been devoted to bridge this gap by building tools for the extraction of high-level, semantics-based features from multimedia content, as low-level features are not considered useful because they deal primarily with representing the perceived content rather than the semantics of it.
In this paper, we explore a different point of view by leveraging the gap between low-level and high-level features. We experiment with a recent approach for movie recommendation that extract low-level Mise-en-Scéne features from multimedia content and combine it with high-level features provided by the wisdom of the crowd.
To this end, we first performed an offline performance assessment by implementing a pure content-based recommender system with three different versions of the same algorithm, respectively based on (i) conventional movie attributes, (ii) mise-en-scene features, and (iii) a hybrid method that interleaves recommendations based on movie attributes and mise-en-scene features. In a second study, we designed an empirical study involving 100 subjects and collected data regarding the quality perceived by the users. Results from both studies show that the introduction of mise-en-scéne features in conjunction with traditional movie attributes improves both offline and online quality of recommendations.},
note = {Pre SFI},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
In the last years, there has been much attention given to the semantic gap problem in multimedia retrieval systems. Much effort has been devoted to bridge this gap by building tools for the extraction of high-level, semantics-based features from multimedia content, as low-level features are not considered useful because they deal primarily with representing the perceived content rather than the semantics of it.
In this paper, we explore a different point of view by leveraging the gap between low-level and high-level features. We experiment with a recent approach for movie recommendation that extract low-level Mise-en-Scéne features from multimedia content and combine it with high-level features provided by the wisdom of the crowd.
To this end, we first performed an offline performance assessment by implementing a pure content-based recommender system with three different versions of the same algorithm, respectively based on (i) conventional movie attributes, (ii) mise-en-scene features, and (iii) a hybrid method that interleaves recommendations based on movie attributes and mise-en-scene features. In a second study, we designed an empirical study involving 100 subjects and collected data regarding the quality perceived by the users. Results from both studies show that the introduction of mise-en-scéne features in conjunction with traditional movie attributes improves both offline and online quality of recommendations.
In this paper, we explore a different point of view by leveraging the gap between low-level and high-level features. We experiment with a recent approach for movie recommendation that extract low-level Mise-en-Scéne features from multimedia content and combine it with high-level features provided by the wisdom of the crowd.
To this end, we first performed an offline performance assessment by implementing a pure content-based recommender system with three different versions of the same algorithm, respectively based on (i) conventional movie attributes, (ii) mise-en-scene features, and (iii) a hybrid method that interleaves recommendations based on movie attributes and mise-en-scene features. In a second study, we designed an empirical study involving 100 subjects and collected data regarding the quality perceived by the users. Results from both studies show that the introduction of mise-en-scéne features in conjunction with traditional movie attributes improves both offline and online quality of recommendations.
Elahi, Mehdi; Hosseini, Reza; Rimaz, Mohammad Hossein; Moghaddam, Farshad B.; Trattner, Christoph
Visually-Aware Video Recommendation in the Cold Start Conference
Proccedings of theACM Hypertext 2020 2017, (Pre SFI).
@conference{Elahi2017b,
title = {Visually-Aware Video Recommendation in the Cold Start},
author = {Mehdi Elahi and Reza Hosseini and Mohammad Hossein Rimaz and Farshad B. Moghaddam and Christoph Trattner},
url = {https://christophtrattner.com/pubs/ht2020.pdf},
year = {2017},
date = {2017-07-01},
urldate = {2017-07-01},
pages = {1-5},
organization = {Proccedings of theACM Hypertext 2020},
abstract = {Recommender Systems (RSs) have become essential tools in any
video-sharing platforms (such as YouTube) by generating video
suggestions for users. Although, RSs have been e!ective, however,
they su!er from the so-called New Item problem. New item problem,
as part of Cold Start problem, happens when a new item is added to
the system catalogue and the RS has no or little data available for
that new item. In such a case, the system may fail to meaningfully
recommend the new item to any user.
In this paper, we propose a novel recommendation technique
based on visual tags, i.e., tags that are automatically annotated
to videos based on visual description of videos. Such visual tags
can be used in an extreme cold start situation, where neither any
rating, nor any tag is available for the new video item. The visual
tags could also be used in the moderate cold start situation when
the new video item has been annotated with few tags. This type
of content features can be extracted automatically without any
human involvement and have been shown to be very e!ective in
representing the video content.
We have used a large dataset of videos and shown that automatically
extracted visual tags can be incorporated into the cold start
recommendation process and achieve superior results compared to
the recommendation based on human-annotated tags.},
note = {Pre SFI},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Recommender Systems (RSs) have become essential tools in any
video-sharing platforms (such as YouTube) by generating video
suggestions for users. Although, RSs have been e!ective, however,
they su!er from the so-called New Item problem. New item problem,
as part of Cold Start problem, happens when a new item is added to
the system catalogue and the RS has no or little data available for
that new item. In such a case, the system may fail to meaningfully
recommend the new item to any user.
In this paper, we propose a novel recommendation technique
based on visual tags, i.e., tags that are automatically annotated
to videos based on visual description of videos. Such visual tags
can be used in an extreme cold start situation, where neither any
rating, nor any tag is available for the new video item. The visual
tags could also be used in the moderate cold start situation when
the new video item has been annotated with few tags. This type
of content features can be extracted automatically without any
human involvement and have been shown to be very e!ective in
representing the video content.
We have used a large dataset of videos and shown that automatically
extracted visual tags can be incorporated into the cold start
recommendation process and achieve superior results compared to
the recommendation based on human-annotated tags.
video-sharing platforms (such as YouTube) by generating video
suggestions for users. Although, RSs have been e!ective, however,
they su!er from the so-called New Item problem. New item problem,
as part of Cold Start problem, happens when a new item is added to
the system catalogue and the RS has no or little data available for
that new item. In such a case, the system may fail to meaningfully
recommend the new item to any user.
In this paper, we propose a novel recommendation technique
based on visual tags, i.e., tags that are automatically annotated
to videos based on visual description of videos. Such visual tags
can be used in an extreme cold start situation, where neither any
rating, nor any tag is available for the new video item. The visual
tags could also be used in the moderate cold start situation when
the new video item has been annotated with few tags. This type
of content features can be extracted automatically without any
human involvement and have been shown to be very e!ective in
representing the video content.
We have used a large dataset of videos and shown that automatically
extracted visual tags can be incorporated into the cold start
recommendation process and achieve superior results compared to
the recommendation based on human-annotated tags.
2016
Elahi, Mehdi; Ricci, Francesco; Rubens, Neil
A survey of active learning in collaborative filtering recommender systems Journal Article
In: Computer Science Review, vol. 20, pp. 29-50, 2016, (Pre SFI).
@article{Elahi2016,
title = {A survey of active learning in collaborative filtering recommender systems},
author = {Mehdi Elahi and Francesco Ricci and Neil Rubens},
url = {https://reader.elsevier.com/reader/sd/pii/S1574013715300150?token=EA12A462FC07F42733F4F13375217A57D3FDC7F6047C133156CB1F4E4487DF24C5366547DF4530A25942F690233F2E30},
doi = {10.1016/j.cosrev.2016.05.002},
year = {2016},
date = {2016-06-02},
journal = {Computer Science Review},
volume = {20},
pages = {29-50},
abstract = {In collaborative filtering recommender systems user’s preferences are expressed as ratings for items, and each additional rating extends the knowledge of the system and affects the system’s recommendation accuracy. In general, the more ratings are elicited from the users, the more effective the recommendations are. However, the usefulness of each rating may vary significantly, i.e., different ratings may bring a different amount and type of information about the user’s tastes. Hence, specific techniques, which are defined as “active learning strategies”, can be used to selectively choose the items to be presented to the user for rating. In fact, an active learning strategy identifies and adopts criteria for obtaining data that better reflects users’ preferences and enables to generate better recommendations.
So far, a variety of active learning strategies have been proposed in the literature. In this article, we survey recent strategies by grouping them with respect to two distinct dimensions: personalization, i.e., whether the system selected items are different for different users or not, and, hybridization, i.e., whether active learning is guided by a single criterion (heuristic) or by multiple criteria. In addition, we present a comprehensive overview of the evaluation methods and metrics that have been employed by the research community in order to test active learning strategies for collaborative filtering. Finally, we compare the surveyed strategies and provide guidelines for their usage in recommender systems.},
note = {Pre SFI},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In collaborative filtering recommender systems user’s preferences are expressed as ratings for items, and each additional rating extends the knowledge of the system and affects the system’s recommendation accuracy. In general, the more ratings are elicited from the users, the more effective the recommendations are. However, the usefulness of each rating may vary significantly, i.e., different ratings may bring a different amount and type of information about the user’s tastes. Hence, specific techniques, which are defined as “active learning strategies”, can be used to selectively choose the items to be presented to the user for rating. In fact, an active learning strategy identifies and adopts criteria for obtaining data that better reflects users’ preferences and enables to generate better recommendations.
So far, a variety of active learning strategies have been proposed in the literature. In this article, we survey recent strategies by grouping them with respect to two distinct dimensions: personalization, i.e., whether the system selected items are different for different users or not, and, hybridization, i.e., whether active learning is guided by a single criterion (heuristic) or by multiple criteria. In addition, we present a comprehensive overview of the evaluation methods and metrics that have been employed by the research community in order to test active learning strategies for collaborative filtering. Finally, we compare the surveyed strategies and provide guidelines for their usage in recommender systems.
So far, a variety of active learning strategies have been proposed in the literature. In this article, we survey recent strategies by grouping them with respect to two distinct dimensions: personalization, i.e., whether the system selected items are different for different users or not, and, hybridization, i.e., whether active learning is guided by a single criterion (heuristic) or by multiple criteria. In addition, we present a comprehensive overview of the evaluation methods and metrics that have been employed by the research community in order to test active learning strategies for collaborative filtering. Finally, we compare the surveyed strategies and provide guidelines for their usage in recommender systems.
Tobias, Ignacio Fernandez; Braunhofer, Matthias; Elahi, Mehdi; Ricci, Francesco; Cantador, Ivan
Alleviating the new user problem in collaborative filtering by exploiting personality information Journal Article
In: User Modeling and User-Adapted Interaction, vol. 26, no. 2-3, pp. 221-255, 2016, (Pre SFI).
@article{Tobias2016,
title = {Alleviating the new user problem in collaborative filtering by exploiting personality information},
author = {Ignacio Fernandez Tobias and Matthias Braunhofer and Mehdi Elahi and Francesco Ricci and Ivan Cantador},
url = {https://www.researchgate.net/publication/285574429_Alleviating_the_New_User_Problem_in_Collaborative_Filtering_by_Exploiting_Personality_Information},
doi = {10.1007/s11257-016-9172-z},
year = {2016},
date = {2016-06-01},
journal = {User Modeling and User-Adapted Interaction},
volume = {26},
number = {2-3},
pages = {221-255},
abstract = {The new user problem in recommender systems is still challenging, and there is not yet a unique solution that can be applied in any domain or situation. In this paper we analyze viable solutions to the new user problem in collaborative filtering (CF) that are based on the exploitation of user personality information: (a) personality-based CF, which directly improves the recommendation prediction model by incorporating user personality information, (b) personality-based active learning, which utilizes personality information for identifying additional useful preference data in the target recommendation domain to be elicited from the user, and (c) personality-based cross-domain recommendation, which exploits personality information to better use user preference data from auxiliary domains which can be used to compensate the lack of user preference data in the target domain. We benchmark the effectiveness of these methods on large datasets that span several domains, namely movies, music and books. Our results show that personality-aware methods achieve performance improvements that range from 6 to 94 % for users completely new to the system, while increasing the novelty of the recommended items by 3–40 % with respect to the non-personalized popularity baseline. We also discuss the limitations of our approach and the situations in which the proposed methods can be better applied, hence providing guidelines for researchers and practitioners in the field.},
note = {Pre SFI},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The new user problem in recommender systems is still challenging, and there is not yet a unique solution that can be applied in any domain or situation. In this paper we analyze viable solutions to the new user problem in collaborative filtering (CF) that are based on the exploitation of user personality information: (a) personality-based CF, which directly improves the recommendation prediction model by incorporating user personality information, (b) personality-based active learning, which utilizes personality information for identifying additional useful preference data in the target recommendation domain to be elicited from the user, and (c) personality-based cross-domain recommendation, which exploits personality information to better use user preference data from auxiliary domains which can be used to compensate the lack of user preference data in the target domain. We benchmark the effectiveness of these methods on large datasets that span several domains, namely movies, music and books. Our results show that personality-aware methods achieve performance improvements that range from 6 to 94 % for users completely new to the system, while increasing the novelty of the recommended items by 3–40 % with respect to the non-personalized popularity baseline. We also discuss the limitations of our approach and the situations in which the proposed methods can be better applied, hence providing guidelines for researchers and practitioners in the field.
2015
Rubens, Neil; Elahi, Mehdi; Sugiyama, Masashi; Kaplan, Dain
Active learning in recommender systems Book Chapter
In: Ricci, Francesco; Rokach, Lior; Shapira, Bracha (Ed.): pp. 809-846, Springer, 2015, ISBN: 978-1-4899-7637-6, (Pre SFI).
@inbook{Rubens2015,
title = {Active learning in recommender systems},
author = {Neil Rubens and Mehdi Elahi and Masashi Sugiyama and Dain Kaplan},
editor = {Francesco Ricci and Lior Rokach and Bracha Shapira},
url = {https://link.springer.com/chapter/10.1007/978-1-4899-7637-6_24},
doi = {10.1007/978-1-4899-7637-6_24},
isbn = {978-1-4899-7637-6},
year = {2015},
date = {2015-01-01},
pages = {809-846},
publisher = {Springer},
abstract = {In Recommender Systems (RS), a user’s preferences are expressed in terms of rated items, where incorporating each rating may improve the RS’s predictive accuracy. In addition to a user rating items at-will (a passive process), RSs may also actively elicit the user to rate items, a process known as Active Learning (AL). However, the number of interactions between the RS and the user is still limited. One aim of AL is therefore the selection of items whose ratings are likely to provide the most information about the user’s preferences. In this chapter, we provide an overview of AL within RSs, discuss general objectives and considerations, and then summarize a variety of methods commonly employed. AL methods are categorized based on our interpretation of their primary motivation/goal, and then sub-classified into two commonly classified types, instance-based and model-based, for easier comprehension. We conclude the chapter by outlining ways in which AL methods could be evaluated, and provide a brief summary of methods performance.},
note = {Pre SFI},
keywords = {},
pubstate = {published},
tppubtype = {inbook}
}
In Recommender Systems (RS), a user’s preferences are expressed in terms of rated items, where incorporating each rating may improve the RS’s predictive accuracy. In addition to a user rating items at-will (a passive process), RSs may also actively elicit the user to rate items, a process known as Active Learning (AL). However, the number of interactions between the RS and the user is still limited. One aim of AL is therefore the selection of items whose ratings are likely to provide the most information about the user’s preferences. In this chapter, we provide an overview of AL within RSs, discuss general objectives and considerations, and then summarize a variety of methods commonly employed. AL methods are categorized based on our interpretation of their primary motivation/goal, and then sub-classified into two commonly classified types, instance-based and model-based, for easier comprehension. We conclude the chapter by outlining ways in which AL methods could be evaluated, and provide a brief summary of methods performance.
0000
Abdollahpouri, Himan; Elahi, Mehdi; Mansoury, Masoud; Sahebi, Shaghayegh; Nazari, Zahra; Chaney, Allison; Loni, Babak
MORS 2021: 1st Workshop on Multi-Objective Recommender Systems. Proceedings
0000.
@proceedings{Abdollahpouri2021,
title = {MORS 2021: 1st Workshop on Multi-Objective Recommender Systems.},
author = {Himan Abdollahpouri and Mehdi Elahi and Masoud Mansoury and Shaghayegh Sahebi and Zahra Nazari and Allison Chaney and Babak Loni},
journal = {In Fifteenth ACM Conference on Recommender Systems.},
pages = {787-788},
keywords = {},
pubstate = {published},
tppubtype = {proceedings}
}