targz

Fairness and relevance may be diverging goals in recommender systems. Can we find a way to achieve both? We talk about it with Dr. Theresia Veronika Rampisela in this episode of targz!Do you want to dig more? Checkout the paper: "Joint Evaluation of Fairness and Relevance in Recommender Systems with Pareto Frontier"

In this episode of targz, Franco Maria Nardini, Research Director at ISTI-CNR, explains Seismic, a two-level inverted index for fast retrieval over learned sparse representations. It beats graph-based state-of-the-art methods by up to 3.5x in speed with comparable memory, and opens new directions in inference-free and edge retrieval.Want to know more? checkout the paper: Efficient Inverted Indexes for Approximate Retrieval over Learned Sparse Representations

Everyone would like a fair recruitment process, but unfortunately the reality is way more complex than just fixing some sorting algorithm. In this episode of targz Dr. Carlos Castillo, aka ChaTo, from ICREA describes the research conducted by his group to address the issue.Want to know more? checkout the paper: https://www.sciencedirect.com/science/article/pii/S0306457325004479Here you can also find more information about the project: http://findhr.eu/

In this episode of targz Marco Minici, Researcher at ICAR-CNR, describes how to identify group of users coordinating on different social platform that try to influence other people opinions.Link to the paper (Exposing Cross-Platform Coordinated Inauthentic Activity in the Run-Up to the 2024 U.S. Election): https://arxiv.org/pdf/2410.22716v3If you want to keep up with every new episode of targz, follow me on:LikedIn: https://www.linkedin.com/in/elleflorio/Bluesky: https://bsky.app/profile/florio.dev

When it comes to recommendation, indirect feedback by user is a powerful tool, but it can be problematic to deal with noise. In this episode of targz Ervin Dervishaj from University of Copenhagen presents a method to leverage LLMs for post-training denoising. How does it work? What are the benefits? Let's find out together!Link to the paper (Post-Training Denoising of User Profiles withLLMs in Collaborative Filtering Recommendation): https://arxiv.org/pdf/2601.18009If you want to keep up with every new episode of targz, follow me on:- LikedIn: https://www.linkedin.com/in/elleflorio/- Bluesky: https://bsky.app/profile/florio.dev

With Giovanni Mauro from Scuola Normale Superiore and Istituto di scienza e tecnologie dell'informazione "A. Faedo" (Cnr-Isti) we discuss how AI influences users and how users influence AI back and the urba`n impact of this feedback loop. Don't miss this third episode of targz!Link to the paper: https://link.springer.com/article/10.1007/s10994-025-06904-zIf you want to keep up with every new episode of targz, follow me on:- LikedIn: https://www.linkedin.com/in/elleflorio/- Bluesky: https://bsky.app/profile/florio.dev

In this second episode of targz we talk about intrinsic dimension in neural networks and why they it is important to understand what's happening inside them. Also, do you know what a manifold is?Link to the paper: https://arxiv.org/abs/1905.12784If you want to keep up with every new episode of targz, follow me on:- LikedIn: https://www.linkedin.com/in/elleflorio/- Bluesky: https://bsky.app/profile/florio.dev

Is it possible that adding random documents to an LLM prompt can actually improve the response? Well, let's find it out in this first episode of targz!Link to the paper: https://arxiv.org/pdf/2401.14887If you want to keep up with every new episode of targz, follow me on:- LikedIn: https://www.linkedin.com/in/elleflorio/- Bluesky: https://bsky.app/profile/florio.dev