Francesco Fabbri
Research Subjects
Francesco Fabbri has a total of 112 co-authors. Here are the top co-authors by number of shared publications:
- Ludovico Boratto
University of Cagliari, Cagliari, Italy
7 shared publications - Giacomo Medda
University of Cagliari, Cagliari, Italy
6 shared publications - Gianni Fenu
University of Cagliari, Cagliari, Italy
6 shared publications - Mirko Marras
University of Cagliari, Cagliari, Italy
6 shared publications - Mounia Lalmas
Spotify, London, United Kingdom
6 shared publications - Andreas Damianou
Spotify, London, United Kingdom
5 shared publications
Latest Publications
- Open Access
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 7141-7150
From Habits to Discovery: Deploying LLMs for Personalized Generative Recommendations at Spotify
Podcast listening is often grounded in a set of favorite shows, while listener intent can evolve over time. This combination of stable preferences and changing intent motivates recommendation approaches that support both familiarity and exploration. Traditional recommender systems typically emphasize long-term interaction patterns, and are less explicitly designed to incorporate rich contextual signals or flexible, intent-aware discovery objectives. In this setting, models that can jointly reason over semantics, context, and user state offer a promising direction. Large Language Models (LLMs) provide strong semantic reasoning and contextual conditioning for discovery-oriented recommendation, but deploying them in production introduces challenges in catalog grounding, user-level personalization, and latency-critical serving. We address these challenges with GLIDE, a production-scale generative recommender for podcast discovery at Spotify. GLIDE formulates recommendation as an instruction-following task over a discretized catalog using Semantic IDs, enabling grounded generation over a large inventory. The model conditions on recent listening history and lightweight user context, while injecting long-term user embeddings as soft prompts to capture stable preferences under strict inference constraints. We evaluate GLIDE using offline retrieval metrics, human judgments, and LLM-based evaluation, and validate its impact through large-scale online A/B testing. Across experiments involving millions of users, GLIDE increases non-habitual podcast streaming on Spotify home surface by up to 5.4% and new-show discovery by up to 14.3%, while meeting production cost and latency constraints.
- 8 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 7141-7150
Open AccessFrom Habits to Discovery: Deploying LLMs for Personalized Generative Recommendations at Spotify
Podcast listening is often grounded in a set of favorite shows, while listener intent can evolve over time. This combination of stable preferences and changing intent motivates recommendation approaches that support both familiarity and exploration. Traditional recommender systems typically emphasize long-term interaction patterns, and are less explicitly designed to incorporate rich contextual signals or flexible, intent-aware discovery objectives. In this setting, models that can jointly reason over semantics, context, and user state offer a promising direction. Large Language Models (LLMs) provide strong semantic reasoning and contextual conditioning for discovery-oriented recommendation, but deploying them in production introduces challenges in catalog grounding, user-level personalization, and latency-critical serving. We address these challenges with GLIDE, a production-scale generative recommender for podcast discovery at Spotify. GLIDE formulates recommendation as an instruction-following task over a discretized catalog using Semantic IDs, enabling grounded generation over a large inventory. The model conditions on recent listening history and lightweight user context, while injecting long-term user embeddings as soft prompts to capture stable preferences under strict inference constraints. We evaluate GLIDE using offline retrieval metrics, human judgments, and LLM-based evaluation, and validate its impact through large-scale online A/B testing. Across experiments involving millions of users, GLIDE increases non-habitual podcast streaming on Spotify home surface by up to 5.4% and new-show discovery by up to 14.3%, while meeting production cost and latency constraints.
- Open Access
- 20 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Recommender Systems
- Vol. 4 (4) , 1-27
Graph Augmentation for Intersectional Unfairness Mitigation: A Study across Dataset Scales and Interaction Densities
Recent work on fairness-aware graph collaborative filtering (GCF) has shown the effectiveness of graph augmentation as a post-processing strategy for mitigating consumer unfairness. However, most studies remain confined to binary fairness setups and operate under limited experimental conditions, often relying on sparse and small-scale datasets. In this paper, we extend our fairness-aware augmentation method to address intersectional unfairness across demographic subgroups, a setting where the intersection of multiple sensitive attributes leads to fine-grained subgroups. To this end, we reformulate the fairness objective to incorporate intersectional demographic groups and evaluate our extended method across interaction configurations that vary in density and scale. Our results reveal that the effectiveness of fair graph augmentations is model-dependent and sensitive to dataset properties. We show that the edges selected during augmentation tend to concentrate around interpretable structural patterns driven by the connected nodes’ characteristics. Furthermore, analyzing how these augmented edges differ across graph-level attributes offers actionable insights into the potential benefits of fairness-oriented graph modifications. Finally, we compare our method with recent fairness-aware baselines, explore the impact of augmenting different graph regions, and assess our mitigation strategy under scenarios with minimal unfairness. Source code publicly available at: https://github.com/jackmedda/Intersectional-FA4GCF.
- 20 February 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Recommender Systems
- Vol. 4 (4) , 1-27
Open AccessGraph Augmentation for Intersectional Unfairness Mitigation: A Study across Dataset Scales and Interaction Densities
Recent work on fairness-aware graph collaborative filtering (GCF) has shown the effectiveness of graph augmentation as a post-processing strategy for mitigating consumer unfairness. However, most studies remain confined to binary fairness setups and operate under limited experimental conditions, often relying on sparse and small-scale datasets. In this paper, we extend our fairness-aware augmentation method to address intersectional unfairness across demographic subgroups, a setting where the intersection of multiple sensitive attributes leads to fine-grained subgroups. To this end, we reformulate the fairness objective to incorporate intersectional demographic groups and evaluate our extended method across interaction configurations that vary in density and scale. Our results reveal that the effectiveness of fair graph augmentations is model-dependent and sensitive to dataset properties. We show that the edges selected during augmentation tend to concentrate around interpretable structural patterns driven by the connected nodes’ characteristics. Furthermore, analyzing how these augmented edges differ across graph-level attributes offers actionable insights into the potential benefits of fairness-oriented graph modifications. Finally, we compare our method with recent fairness-aware baselines, explore the impact of augmenting different graph regions, and assess our mitigation strategy under scenarios with minimal unfairness. Source code publicly available at: https://github.com/jackmedda/Intersectional-FA4GCF.
- Open Access
Do LLM-judges Align with Human Relevance in Cranfield-style Recommender Evaluation?
Evaluating recommender systems remains a long-standing challenge, as offline methods based on historical user interactions and train-test splits often yield unstable and inconsistent results due to exposure bias, popularity bias, sampled evaluations, and missing-not-at-random patterns. In contrast, textual document retrieval benefits from robust, standardized evaluation via Cranfield-style test collections, which combine pooled relevance judgments with controlled setups. While recent work shows that adapting this methodology to recommender systems is feasible, constructing such collections remains costly due to the need for manual relevance judgments, thus limiting scalability. This paper investigates whether Large Language Models (LLMs) can serve as reliable automatic judges to address these scalability challenges. Using the ML-32M-ext Cranfield-style movie recommendation collection, we first examine the limitations of existing evaluation methodologies. Then we explore the alignment and the recommender systems ranking agreement between the LLM-judge and human provided relevance labels. We find that incorporating richer item metadata and longer user histories improves alignment, and that LLM-judge yields high agreement with human-based rankings (Kendall's tau = 0.87). Finally, an industrial case study in the podcast recommendation domain demonstrates the practical value of LLM-judge for model selection. Overall, our results show that LLM-judge is a viable and scalable approach for evaluating recommender systems.
Open AccessDo LLM-judges Align with Human Relevance in Cranfield-style Recommender Evaluation?
Evaluating recommender systems remains a long-standing challenge, as offline methods based on historical user interactions and train-test splits often yield unstable and inconsistent results due to exposure bias, popularity bias, sampled evaluations, and missing-not-at-random patterns. In contrast, textual document retrieval benefits from robust, standardized evaluation via Cranfield-style test collections, which combine pooled relevance judgments with controlled setups. While recent work shows that adapting this methodology to recommender systems is feasible, constructing such collections remains costly due to the need for manual relevance judgments, thus limiting scalability. This paper investigates whether Large Language Models (LLMs) can serve as reliable automatic judges to address these scalability challenges. Using the ML-32M-ext Cranfield-style movie recommendation collection, we first examine the limitations of existing evaluation methodologies. Then we explore the alignment and the recommender systems ranking agreement between the LLM-judge and human provided relevance labels. We find that incorporating richer item metadata and longer user histories improves alignment, and that LLM-judge yields high agreement with human-based rankings (Kendall's tau = 0.87). Finally, an industrial case study in the podcast recommendation domain demonstrates the practical value of LLM-judge for model selection. Overall, our results show that LLM-judge is a viable and scalable approach for evaluating recommender systems.
- Open Access
- 22 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 401-410
Prompt-to-Slate: Diffusion Models for Prompt-Conditioned Slate Generation
- 22 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 401-410
Open AccessPrompt-to-Slate: Diffusion Models for Prompt-Conditioned Slate Generation
- Top Cited PapersOpen Access
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 840-854
Policy-as-Prompt: Rethinking Content Moderation in the Age of Large Language Models
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 840-854
Top Cited PapersOpen AccessPolicy-as-Prompt: Rethinking Content Moderation in the Age of Large Language Models
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (27) , 28258-28266
IOHunter: Graph Foundation Model to Uncover Online Information Operations
Social media platforms have become vital spaces for public discourse, serving as modern agorás where a wide range of voices influence societal narratives. However, their open nature also makes them vulnerable to exploitation by malicious actors, including state-sponsored entities, who can conduct information operations (IOs) to manipulate public opinion. The spread of misinformation, false news, and misleading claims threatens democratic processes and societal cohesion, making it crucial to develop methods for the timely detection of inauthentic activity to protect the integrity of online discourse. In this work, we introduce a methodology designed to identify users orchestrating information operations, a.k.a. IO drivers, across various influence campaigns. Our framework, named IOHunter, leverages the combined strengths of Language Models and Graph Neural Networks to improve generalization in supervised, scarcely-supervised, and cross-IO contexts. Our approach achieves state-of-the-art performance across multiple sets of IOs originating from six countries, significantly surpassing existing approaches. This research marks a step toward developing Graph Foundation Models specifically tailored for the task of IO detection on social media platforms.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (27) , 28258-28266
Open AccessIOHunter: Graph Foundation Model to Uncover Online Information Operations
Social media platforms have become vital spaces for public discourse, serving as modern agorás where a wide range of voices influence societal narratives. However, their open nature also makes them vulnerable to exploitation by malicious actors, including state-sponsored entities, who can conduct information operations (IOs) to manipulate public opinion. The spread of misinformation, false news, and misleading claims threatens democratic processes and societal cohesion, making it crucial to develop methods for the timely detection of inauthentic activity to protect the integrity of online discourse. In this work, we introduce a methodology designed to identify users orchestrating information operations, a.k.a. IO drivers, across various influence campaigns. Our framework, named IOHunter, leverages the combined strengths of Language Models and Graph Neural Networks to improve generalization in supervised, scarcely-supervised, and cross-IO contexts. Our approach achieves state-of-the-art performance across multiple sets of IOs originating from six countries, significantly surpassing existing approaches. This research marks a step toward developing Graph Foundation Models specifically tailored for the task of IO detection on social media platforms.
- Open Access
- 27 March 2025
- Published by Elsevier in Information Processing & Management
- Vol. 62 (4)
Algorithmic Drift: A simulation framework to study the effects of recommender systems on user preferences
- 27 March 2025
- Published by Elsevier in Information Processing & Management
- Vol. 62 (4)
Open AccessAlgorithmic Drift: A simulation framework to study the effects of recommender systems on user preferences
- Open Access
Policy-as-Prompt: Rethinking Content Moderation in the Age of Large Language Models
Content moderation plays a critical role in shaping safe and inclusive online environments, balancing platform standards, user expectations, and regulatory frameworks. Traditionally, this process involves operationalising policies into guidelines, which are then used by downstream human moderators for enforcement, or to further annotate datasets for training machine learning moderation models. However, recent advancements in large language models (LLMs) are transforming this landscape. These models can now interpret policies directly as textual inputs, eliminating the need for extensive data curation. This approach offers unprecedented flexibility, as moderation can be dynamically adjusted through natural language interactions. This paradigm shift raises important questions about how policies are operationalised and the implications for content moderation practices. In this paper, we formalise the emerging policy-as-prompt framework and identify five key challenges across four domains: Technical Implementation (1. translating policy to prompts, 2. sensitivity to prompt structure and formatting), Sociotechnical (3. the risk of technological determinism in policy formation), Organisational (4. evolving roles between policy and machine learning teams), and Governance (5. model governance and accountability). Through analysing these challenges across technical, sociotechnical, organisational, and governance dimensions, we discuss potential mitigation approaches. This research provides actionable insights for practitioners and lays the groundwork for future exploration of scalable and adaptive content moderation systems in digital ecosystems.
Open AccessPolicy-as-Prompt: Rethinking Content Moderation in the Age of Large Language Models
Content moderation plays a critical role in shaping safe and inclusive online environments, balancing platform standards, user expectations, and regulatory frameworks. Traditionally, this process involves operationalising policies into guidelines, which are then used by downstream human moderators for enforcement, or to further annotate datasets for training machine learning moderation models. However, recent advancements in large language models (LLMs) are transforming this landscape. These models can now interpret policies directly as textual inputs, eliminating the need for extensive data curation. This approach offers unprecedented flexibility, as moderation can be dynamically adjusted through natural language interactions. This paradigm shift raises important questions about how policies are operationalised and the implications for content moderation practices. In this paper, we formalise the emerging policy-as-prompt framework and identify five key challenges across four domains: Technical Implementation (1. translating policy to prompts, 2. sensitivity to prompt structure and formatting), Sociotechnical (3. the risk of technological determinism in policy formation), Organisational (4. evolving roles between policy and machine learning teams), and Governance (5. model governance and accountability). Through analysing these challenges across technical, sociotechnical, organisational, and governance dimensions, we discuss potential mitigation approaches. This research provides actionable insights for practitioners and lays the groundwork for future exploration of scalable and adaptive content moderation systems in digital ecosystems.
- Open Access
- 26 December 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Intelligent Systems and Technology
- Vol. 16 (1) , 1-26
GNNUERS: Fairness Explanation in GNNs for Recommendation via Counterfactual Reasoning
Nowadays, research into personalization has been focusing on explainability and fairness. Several approaches proposed in recent works are able to explain individual recommendations in a post-hoc manner or by explanation paths. However, explainability techniques applied to unfairness in recommendation have been limited to finding user/item features mostly related to biased recommendations. In this article, we devised a novel algorithm that leverages counterfactuality methods to discover user unfairness explanations in the form of user-item interactions. In our counterfactual framework, interactions are represented as edges in a bipartite graph, with users and items as nodes. Our bipartite graph explainer perturbs the topological structure to find an altered version that minimizes the disparity in utility between the protected and unprotected demographic groups. Experiments on four real-world graphs coming from various domains showed that our method can systematically explain user unfairness on three state-of-the-art GNN-based recommendation models. Moreover, an empirical evaluation of the perturbed network uncovered relevant patterns that justify the nature of the unfairness discovered by the generated explanations. The source code and the preprocessed data sets are available at https://github.com/jackmedda/RS-BGExplainer .
- 26 December 2024
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Intelligent Systems and Technology
- Vol. 16 (1) , 1-26
Open AccessGNNUERS: Fairness Explanation in GNNs for Recommendation via Counterfactual Reasoning
Nowadays, research into personalization has been focusing on explainability and fairness. Several approaches proposed in recent works are able to explain individual recommendations in a post-hoc manner or by explanation paths. However, explainability techniques applied to unfairness in recommendation have been limited to finding user/item features mostly related to biased recommendations. In this article, we devised a novel algorithm that leverages counterfactuality methods to discover user unfairness explanations in the form of user-item interactions. In our counterfactual framework, interactions are represented as edges in a bipartite graph, with users and items as nodes. Our bipartite graph explainer perturbs the topological structure to find an altered version that minimizes the disparity in utility between the protected and unprotected demographic groups. Experiments on four real-world graphs coming from various domains showed that our method can systematically explain user unfairness on three state-of-the-art GNN-based recommendation models. Moreover, an empirical evaluation of the perturbed network uncovered relevant patterns that justify the nature of the unfairness discovered by the generated explanations. The source code and the preprocessed data sets are available at https://github.com/jackmedda/RS-BGExplainer .
- Open Access
- 8 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 158-168
Fair Augmentation for Graph Collaborative Filtering
Recent developments in recommendation have harnessed the collaborative power of graph neural networks (GNNs) in learning users’ preferences from user-item networks. Despite emerging regulations addressing fairness of automated systems, unfairness issues in graph collaborative filtering remain underexplored, especially from the consumer’s perspective. Despite numerous contributions on consumer unfairness, only a few of these works have delved into GNNs. A notable gap exists in the formalization of the latest mitigation algorithms, as well as in their effectiveness and reliability on cutting-edge models. This paper serves as a solid response to recent research highlighting unfairness issues in graph collaborative filtering by reproducing one of the latest mitigation methods. The reproduced technique adjusts the system fairness level by learning a fair graph augmentation. Under an experimental setup based on 11 GNNs, 5 non-GNN models, and 5 real-world networks across diverse domains, our investigation reveals that fair graph augmentation is consistently effective on high-utility models and large datasets. Experiments on the transferability of the fair augmented graph open new issues for future recommendation studies. Source code: https://github.com/jackmedda/FA4GCF.
- 8 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 158-168
Open AccessFair Augmentation for Graph Collaborative Filtering
Recent developments in recommendation have harnessed the collaborative power of graph neural networks (GNNs) in learning users’ preferences from user-item networks. Despite emerging regulations addressing fairness of automated systems, unfairness issues in graph collaborative filtering remain underexplored, especially from the consumer’s perspective. Despite numerous contributions on consumer unfairness, only a few of these works have delved into GNNs. A notable gap exists in the formalization of the latest mitigation algorithms, as well as in their effectiveness and reliability on cutting-edge models. This paper serves as a solid response to recent research highlighting unfairness issues in graph collaborative filtering by reproducing one of the latest mitigation methods. The reproduced technique adjusts the system fairness level by learning a fair graph augmentation. Under an experimental setup based on 11 GNNs, 5 non-GNN models, and 5 real-world networks across diverse domains, our investigation reveals that fair graph augmentation is consistently effective on high-utility models and large datasets. Experiments on the transferability of the fair augmented graph open new issues for future recommendation studies. Source code: https://github.com/jackmedda/FA4GCF.