المؤلفون: Fabbri, Francesco

المساهمون: University/Department: Universitat Pompeu Fabra. Departament de Tecnologies de la Informació i les Comunicacions

مرشدي الرسالة: Bonchi, Francesco

المصدر: TDX (Tesis Doctorals en Xarxa)

مصطلحات موضوعية: Recomender systems, Sistemas de recomendación, Sistemes de recomanació

الوصف: Recommender Systems represent a key instrument to convey consumption of contents available on the Web. They enhance the engagement among the users and the online platforms through algorithmic personalization. Injecting non-natural interactions consequently cannot have only beneficial effects. Indeed, amplifying and exaggerating human behaviors leads to either the spread of extreme point of views (e.g. polarized or controversial opinions) or the discrimination or mistreatment of a specific group of individuals. In this thesis, we pose the attention on the importance of auditing and mitigating the “algorithmic bias” generated by a recommendation system, emphasizing its role on the networked interactions of users and contents. Through empirical evidences we highlight how the social graph, presenting biased network topology, when used as input, can impact the algorithmic recommendations. This analysis allows to add a perspective on the long-term impact of algorithmic suggestions, leading to design a simulation model able to explain the “feedback-loop” generated on social networks. Auditing the algorithmic bias facilitates the design of strategies able to mitigate algorithmic risks in recommendation, such as radicalization and unfairness. The results found in this thesis raise critical observations about the impact of recommendation algorithms, and hints of the need to design systems able to mitigate biases embedded in data and algorithms, considering both short and long-term perspectives.

الوصف (مترجم): Los sistemas de recomendación representan un instrumento clave para vehicular el consumo de contenidos disponibles en la Web. Mejoran el vínculo entre los usuarios y las plataformas en línea a través de la personalización algorítmica. En consecuencia, la inyección de interacciones no naturales no tiene sólo efectos positivos. La amplificación y exageración de los comportamientos humanos conduce a la difusión de puntos de vista extremos (por ejemplo, opiniones polarizadas o controvertidas) y a la discriminación o el maltrato de un grupo específico de individuos. En esta tesis, se pone la atención en la importancia de auditar y mitigar el ”sesgo algorítmico” generado por un sistema de recomendación, enfatizando su función en las interacciones en redes de usuarios y de contenidos. A través de evidencias empíricas evidenciamos cómo el grafo social, que presenta una topología de red sesgada, puede impactar en las recomendaciones algorítmicas, cuando se utiliza como input. Este análisis permite añadir una perspectiva sobre el impacto a largo plazo de las sugerencias algorítmicas, llevando a diseñar un modelo de simulación que permite de explicar el “feedback-loop” por las mismas en las redes sociales. La comprobación del sesgo algorítmico facilita el diseño de estrategias capaces de mitigar los riesgos algorítmicos en la recomendación, como la radicalización y la injusticia. Los resultados obtenidos plantean observaciones críticas sobre el impacto de los algoritmos de recomendación, e insinúan la necesidad de diseñar sistemas capaces de mitigar los sesgos incorporados a los datos y a los algoritmos, teniendo en cuenta tanto las perspectivas a corto como a largo plazo.
Els sistemes de recomanació representen un instrument clau per vehicular el consum de continguts disponibles a la web. Milloren el compromís entre els usuaris i les plataformes en línia mitjançant la personalització algorítmica. Per tant, la injecció d’interaccions no naturals no només té efectes positius. L’amplificació i l’exageració dels comportaments humans condueix a la difusió de punts de vista extrems (per exemple, opinions polaritzades o controvertides) o a la discriminació o el maltractament d’un grup específic d’individus. En aquesta tesi, es posa l’atenció en la importància d’auditar i mitigar el ”biaix algorítmic” generat per un sistema de recomanació, emfatitzant-ne la funció en les interaccions en xarxes d’usuaris i continguts. A través d’evid`encies empíriques evidenciem com el graf social, que presenta una topologia de xarxa esbiaixada, pot impactar en les recomanacions algorítmiques quan s’utilitza com a input. Aquesta anàlisi permet afegir una perspectiva sobre l’impacte a llarg termini dels suggeriments algorítmics, portant a dissenyar un model de simulació que permet explicar el ”feedback-loop” generat per aquestes a les xarxes socials. Aquesta an`alisi va facilitar el disseny d’estratègies capaces de mitigar els riscos algor´ıtmics en la recomanació, com ara la radicalització i la injusíıcia. Els nostres resultats plantegen observacions crítiques sobre l’impacte dels algorismes de recomanació, i insinuen la necessitat de dissenyar sistemes capaços de mitigar els biaixos incorporats a les dades i als algoritmes, considerant tant les perspectives a curt com a llarg termini.
Programa de doctorat en Tecnologies de la Informació i les Comunicacions

وصف الملف: application/pdf

الوصول الحر: http://hdl.handle.net/10803/675938Test

المؤلفون: Fabbri, Francesco, Principi, Giulio, Stanca, Lorenzo

مصطلحات موضوعية: Economics - Theoretical Economics

الوصف: We represent preferences that exhibit absolute or relative attitudes towards ambiguity without assuming convexity of preferences. Our analysis is motivated by the recent experimental evidence by Baillon and Placido (2019) indicating that ambiguity becomes more tolerable as individuals are better off overall. Decreasing absolute ambiguity aversion is characterized by constant superadditive certainty equivalents and admits an act-dependent variational representation (Maccheroni et al., 2006). Decreasing relative ambiguity aversion relates to positive superhomogeneity and admits an act-dependent confidence preference representation (Chateauneuf and Faro, 2009). We apply our characterizations to retrieve a classic risk sharing result on the efficiency of trade and subjective beliefs of the individuals (Rigotti et al., 2008).

الوصول الحر: http://arxiv.org/abs/2406.01343Test

المؤلفون: Damianou, Andreas, Fabbri, Francesco, Gigioli, Paul, De Nadai, Marco, Wang, Alice, Palumbo, Enrico, Lalmas, Mounia

مصطلحات موضوعية: Computer Science - Information Retrieval, Computer Science - Machine Learning

الوصف: In the realm of personalization, integrating diverse information sources such as consumption signals and content-based representations is becoming increasingly critical to build state-of-the-art solutions. In this regard, two of the biggest trends in research around this subject are Graph Neural Networks (GNNs) and Foundation Models (FMs). While GNNs emerged as a popular solution in industry for powering personalization at scale, FMs have only recently caught attention for their promising performance in personalization tasks like ranking and retrieval. In this paper, we present a graph-based foundation modeling approach tailored to personalization. Central to this approach is a Heterogeneous GNN (HGNN) designed to capture multi-hop content and consumption relationships across a range of recommendable item types. To ensure the generality required from a Foundation Model, we employ a Large Language Model (LLM) text-based featurization of nodes that accommodates all item types, and construct the graph using co-interaction signals, which inherently transcend content specificity. To facilitate practical generalization, we further couple the HGNN with an adaptation mechanism based on a two-tower (2T) architecture, which also operates agnostically to content type. This multi-stage approach ensures high scalability; while the HGNN produces general purpose embeddings, the 2T component models in a continuous space the sheer size of user-item interaction data. Our comprehensive approach has been rigorously tested and proven effective in delivering recommendations across a diverse array of products within a real-world, industrial audio streaming platform.

الوصول الحر: http://arxiv.org/abs/2403.07478Test

المؤلفون: De Nadai, Marco, Fabbri, Francesco, Gigioli, Paul, Wang, Alice, Li, Ang, Silvestri, Fabrizio, Kim, Laura, Lin, Shawn, Radosavljevic, Vladan, Ghael, Sandeep, Nyhan, David, Bouchard, Hugues, Lalmas-Roelleke, Mounia, Damianou, Andreas

مصطلحات موضوعية: Computer Science - Information Retrieval, Computer Science - Machine Learning

الوصف: In the ever-evolving digital audio landscape, Spotify, well-known for its music and talk content, has recently introduced audiobooks to its vast user base. While promising, this move presents significant challenges for personalized recommendations. Unlike music and podcasts, audiobooks, initially available for a fee, cannot be easily skimmed before purchase, posing higher stakes for the relevance of recommendations. Furthermore, introducing a new content type into an existing platform confronts extreme data sparsity, as most users are unfamiliar with this new content type. Lastly, recommending content to millions of users requires the model to react fast and be scalable. To address these challenges, we leverage podcast and music user preferences and introduce 2T-HGNN, a scalable recommendation system comprising Heterogeneous Graph Neural Networks (HGNNs) and a Two Tower (2T) model. This novel approach uncovers nuanced item relationships while ensuring low latency and complexity. We decouple users from the HGNN graph and propose an innovative multi-link neighbor sampler. These choices, together with the 2T component, significantly reduce the complexity of the HGNN model. Empirical evaluations involving millions of users show significant improvement in the quality of personalized recommendations, resulting in a +46% increase in new audiobooks start rate and a +23% boost in streaming rates. Intriguingly, our model's impact extends beyond audiobooks, benefiting established products like podcasts.
Comment: To appear in The Web Conference 2024 proceedings

الوصول الحر: http://arxiv.org/abs/2403.05185Test

المؤلفون: Boratto, Ludovico, Fabbri, Francesco, Fenu, Gianni, Marras, Mirko, Medda, Giacomo

مصطلحات موضوعية: Computer Science - Information Retrieval

الوصف: Efforts in the recommendation community are shifting from the sole emphasis on utility to considering beyond-utility factors, such as fairness and robustness. Robustness of recommendation models is typically linked to their ability to maintain the original utility when subjected to attacks. Limited research has explored the robustness of a recommendation model in terms of fairness, e.g., the parity in performance across groups, under attack scenarios. In this paper, we aim to assess the robustness of graph-based recommender systems concerning fairness, when exposed to attacks based on edge-level perturbations. To this end, we considered four different fairness operationalizations, including both consumer and provider perspectives. Experiments on three datasets shed light on the impact of perturbations on the targeted fairness notion, uncovering key shortcomings in existing evaluation protocols for robustness. As an example, we observed perturbations affect consumer fairness on a higher extent than provider fairness, with alarming unfairness for the former. Source code: https://github.com/jackmedda/CPFairRobustTest

الوصول الحر: http://arxiv.org/abs/2401.13823Test

المؤلفون: Fabbri, Francesco, Liu, Xianghang, McKenzie, Jack R., Twardowski, Bartlomiej, Wijaya, Tri Kurniawan

مصطلحات موضوعية: Computer Science - Information Retrieval, Computer Science - Machine Learning

الوصف: Federated Learning (FL) has emerged as a key approach for distributed machine learning, enhancing online personalization while ensuring user data privacy. Instead of sending private data to a central server as in traditional approaches, FL decentralizes computations: devices train locally and share updates with a global server. A primary challenge in this setting is achieving fast and accurate model training - vital for recommendation systems where delays can compromise user engagement. This paper introduces FedFNN, an algorithm that accelerates decentralized model training. In FL, only a subset of users are involved in each training epoch. FedFNN employs supervised learning to predict weight updates from unsampled users, using updates from the sampled set. Our evaluations, using real and synthetic data, show: 1. FedFNN achieves training speeds 5x faster than leading methods, maintaining or improving accuracy; 2. the algorithm's performance is consistent regardless of client cluster variations; 3. FedFNN outperforms other methods in scenarios with limited client availability, converging more quickly.

الوصول الحر: http://arxiv.org/abs/2309.08635Test

المؤلفون: Boratto, Ludovico, Fabbri, Francesco, Fenu, Gianni, Marras, Mirko, Medda, Giacomo

مصطلحات موضوعية: Computer Science - Information Retrieval

الوصف: In recommendation literature, explainability and fairness are becoming two prominent perspectives to consider. However, prior works have mostly addressed them separately, for instance by explaining to consumers why a certain item was recommended or mitigating disparate impacts in recommendation utility. None of them has leveraged explainability techniques to inform unfairness mitigation. In this paper, we propose an approach that relies on counterfactual explanations to augment the set of user-item interactions, such that using them while inferring recommendations leads to fairer outcomes. Modeling user-item interactions as a bipartite graph, our approach augments the latter by identifying new user-item edges that not only can explain the original unfairness by design, but can also mitigate it. Experiments on two public data sets show that our approach effectively leads to a better trade-off between fairness and recommendation utility compared with state-of-the-art mitigation procedures. We further analyze the characteristics of added edges to highlight key unfairness patterns. Source code available at https://github.com/jackmedda/RS-BGExplainer/tree/cikm2023Test.
Comment: Accepted as a short paper at CIKM 2023

الوصول الحر: http://arxiv.org/abs/2308.12083Test

المؤلفون: Gamba, Rossella, Breschi, Matteo, Bernuzzi, Sebastiano, Nagar, Alessandro, Cook, William, Doulis, Georgios, Fabbri, Francesco, Ortiz, Néstor, Poudel, Amit, Rashti, Alireza, Tichy, Wolfgang, Ujevic, Maximiliano

مصطلحات موضوعية: General Relativity and Quantum Cosmology, Astrophysics - High Energy Astrophysical Phenomena

الوصف: Gravitational wave astronomy pipelines rely on template waveform models for searches and parameter estimation purposes. For coalescing binary neutron stars (BNS), such models need to accurately reproduce numerical relativity (NR) up to merger, in order to provide robust estimate of the stars' equation of state - dependent parameters. In this work we present an improved version of the Effective One Body (EOB) model $\tt TEOBResumS$ for gravitational waves from BNS systems. Building upon recent post-Newtonian calculations, we include subleading order tidal terms in the waveform multipoles and EOB metric potentials, as well as add up to 5.5PN terms in the gyro-gravitomagnetic functions entering the spin-orbit sector of the model. In order to further improve the EOB-NR agreement in the last few orbital cycles before merger, we introduce next-to-quasicircular corrections in the waveform -- informed by a large number of BNS NR simulations -- and introduce a new NR-informed parameter entering the tidal sector of our conservative dynamics. The performance of our model is then validated against 14 new eccentricity reduced simulations of unequal mass, spinning binaries with varying equation of state. A time-domain phasing analysis and mismatch computations demonstrate that the new model overall improves over the previous version of $\tt TEOBResumS$. Finally, we present a closed-form frequency domain representation of the (tidal) amplitude and phase of the new model. This representation accounts for mass-ratio, aligned spin and (resummed) spin-quadrupole effects in the tidal phase and -- within the calibration region -- it is faithful to the original model.
Comment: 17 pages, 6 figures

الوصول الحر: http://arxiv.org/abs/2307.15125Test

المؤلفون: Medda, Giacomo, Fabbri, Francesco, Marras, Mirko, Boratto, Ludovico, Fenu, Gianni

مصطلحات موضوعية: Computer Science - Information Retrieval

الوصف: 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 paper, 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-BGExplainerTest.

الوصول الحر: http://arxiv.org/abs/2304.06182Test

المؤلفون: La Gatta, Valerio, Luceri, Luca, Fabbri, Francesco, Ferrara, Emilio

مصطلحات موضوعية: Computer Science - Social and Information Networks, H.3.3

الوصف: The proliferation of harmful content shared online poses a threat to online information integrity and the integrity of discussion across platforms. Despite various moderation interventions adopted by social media platforms, researchers and policymakers are calling for holistic solutions. This study explores how a target platform could leverage content that has been deemed harmful on a source platform by investigating the behavior and characteristics of Twitter users responsible for sharing moderated YouTube videos. Using a large-scale dataset of 600M tweets related to the 2020 U.S. election, we find that moderated Youtube videos are extensively shared on Twitter and that users who share these videos also endorse extreme and conspiratorial ideologies. A fraction of these users are eventually suspended by Twitter, but they do not appear to be involved in state-backed information operations. The findings of this study highlight the complex and interconnected nature of harmful cross-platform information diffusion, raising the need for cross-platform moderation strategies.
Comment: 14 pages, 8 figures

الوصول الحر: http://arxiv.org/abs/2304.01371Test