HAL Paris Dauphine-PSL
Not a member yet
21932 research outputs found
Sort by
Kayfabe and the politics of unreality: from Jára Cimrman to Goncharov
No full textFrom the role of kayfabe in wrestling to fictional Czech polymath Jára Cimrman, many communities play around fictional cultural elements whose consumption and diffusion relies on asserting their reality and authenticity, with the expectation that the in-group is wholly aware of the joke. Despite the increased difficulty inherent to the switch to diffuse and fragmented online communities, mystifications keep being created, with memes like Goncharov (a movie that never existed) spreading on platforms like Tumblr, where they can reinforce the feeling of community. A relatively new phenomenon can be observed, partially linked to this meme: the #unreality tag. Often intended as an accessibility feature, its stated purpose is not to explain the meme to those in the out-group, but to help some members of the community who struggle to differentiate reality from their own imagination. In this way, the very meme that would induce extra struggle serves as a way to start conversations on less well-known aspects of neurodivergence and Mad accessibility, such as those concerning psychosis. This article looks at the conditions that allowed this evolution and how it fits in the changing fights for disability rights
The secret behind Temu’s rock-bottom prices
No full textTemu has made a remarkable entry in the global e-commerce landscape, quickly becoming the fifth largest online marketplace worlwide. Critics claim Temu’s ultra-competitive pricing relies on unfair practices. Yet its success stems from the powerful—and proven—business model of its parent company, Pinduoduo, which started as an online marketplace for fresh fruit
L'Intelligence artificielle numérique face au défi des connaissances tacites humaines
No full textNational audienc
Argumentation meets matrix factorization: A dual perspective for explainable recommendations
No full textInternational audienceFactorization-based models gained prominence during the Netflix Challenge (2007) and have since demonstrated strong performance in predicting user ratings. However, their limited interpretability often hinders users from understanding the rationale behind recommendations. In contrast, argumentation-based methods offer a different perspective: they model human-like reasoning by structuring information as arguments and counterarguments. They excel in explainability but typically fall short in accuracy To address this trade-off, we propose a novel framework, Context-Aware Feature-Attribution Through Argumentation (CA-FATA), which combines the predictive power of matrix factorization with the interpretability of argumentation frameworks. In CA-FATA, each user–item interaction is modeled using an argumentation framework. Items’ features are represented as arguments, and users’ ratings determine the arguments’ strengths. Additionally, the model incorporates users’ contextual information (e.g., time, location) to further improve predictive performance. Empirical evaluations on real-world datasets show that CA-FATA excels in predictive accuracy and interpretability. It outperforms existing argumentation-based methods and achieves comparable results with state-of-the-art context-free and context-aware models. CA-FATA also supports multiple explanation formats, including template-based explanations, interactive feedback, and contrastive reasoning. Furthermore, it alleviates the cold-start problem by clustering users based on feature preferences
Am I Secure by Design ?” Evaluating the Security and Transparency of GenAI: An End User-Centric Approach
No full textGenerative AI models, such as ChatGPT and Deepseek, are increasingly integrated into daily life. However, concerns about reliability, cybersecurity, transparency, and data privacy are growing through their use. While Secure by Design (SbD) and Explainable AI (XAI) offer theoretical guidelines, their practical combined application to AI-generated content remains unclear. This study empirically evaluates AI security, cybersecurity and transparency using a structured interrogation method directly addressed to AI models. We assessed multiple text-based open-source and proprietary AI systems on cybersecurity claims, update transparency, and privacy compliance. Preliminary results reveal some discrepancies between AI declarations and actual adherence to SbD principles. While most models incorporate ethical safeguards, they lack clarity on security updates and data management, particularly regarding training data. We propose a user-centered audit framework to test transparency and AI security commitments. The findings emphasize the need to adapt current Secure by Design standards to AI ecosystems while ensuring verifiable transparency
Autonomie, dépendance ou concurrence ? Logiques de structuration des positionnements syndicaux sur la « transition énergétique » à l’échelle d’une raffinerie pétrolière
No full text
