Topic modeling in Italian legal research is hindered by the lack of public datasets, limiting the analysis of legal themes in Supreme Court judgments. To address this, we developed a document processing pipeline that produces an anonymized dataset optimized for topic modeling. The pipeline integrates document layout analysis (YOLOv8x), optical character recognition, and text anonymization. The DLA module achieved a mAP@50 of 0.964 and a mAP@50-95 of 0.800. The OCR detector reached a mAP@50-95 of 0.9022, and the text recognizer (TrOCR) obtained a character error rate of 0.0047 and a word error rate of 0.0248. Compared to OCR-only methods, our dataset improved topic modeling with a diversity score of 0.6198 and a coherence score of 0.6638. We applied BERTopic to extract topics and used large language models to generate labels and summaries. Outputs were evaluated against domain expert interpretations. Claude Sonnet 3.7 achieved a BERTScore F1 of 0.8119 for labeling and 0.9130 for summarization.

Misinformation spans various domains, but detection methods trained on specific domains often perform poorly when applied to others. With the rapid development of Large Language Models (LLMs), researchers have begun to utilize LLMs for cross-domain misinformation detection. However, existing LLM-based methods often fail to adequately analyze news in the target domain, limiting their detection capabilities. More importantly, these methods typically rely on manually designed decision rules, which are limited by domain knowledge and expert experience, thus limiting the generalizability of decision rules to different domains. To address these issues, we propose a MultiAgent Framework for cross-domain misinformation detection with Automated Decision Rule Optimization (MARO). Under this framework, we first employs multiple expert agents to analyze target-domain news. Subsequently, we introduce a question-reflection mechanism that guides expert agents to facilitate higherquality analysis. Furthermore, we propose a decision rule optimization approach based on carefully-designed cross-domain validation tasks to iteratively enhance the effectiveness of decision rules in different domains. Experimental results and in-depth analysis on commonlyused datasets demonstrate that MARO achieves significant improvements over existing methods.

DeepSeek's cut-price challenge to US AI dominance, which wiped 600bn in Nvidia's market value on Monday and caused the tech-weighted Nasdaq index to drop 3%, also took a bite out of the fortunes of some of the world's wealthiest men. Nvidia's record stock plunge, judged to be the biggest market value drop in US stock market history, according to Bloomberg, took with it 20.7bn of its CEO and biggest individual shareholder, Jensen Huang. The share-price drop left Huang, the company co-founder, with a net worth of 103.7bn late on Monday, down from 124.4bn. According to Forbes' real-time billionaires ranking, that pushes the 61-year-old tech tycoon from 10th to 17th place in global wealth rankings, behind Zara fashion mogul Amancio Ortega; Walmart heirs Rob, Jim and Alice Walton; Microsoft cofounder Bill Gates; Dell CEO Michael Dell; and former New York mayor Michael Bloomberg. Exceeding Huang's paper losses from Monday's trading is cat rancher and Oracle chair Larry Ellison, who recorded a 27.6bn loss after shares in Oracle stock dropped 14%, knocking him down from third wealthiest man to fifth, behind Meta CEO Mark Zuckerberg and LVMH luxury tycoon Bernard Arnault.

CES 2025 is here, and that means a whole fleet of new gadgets has been introduced to the world. As we do annually, the Engadget team has battled jet lag, sleep deprivation and the static shocks of those horrible casino carpets to bring you all of the most pressing news coming out of the show this year. You can follow along with our CES liveblog here, but if you don't want to keep a browser tab locked on the site, here's a handy recap for you. Below may not be everything we covered, but it's a rundown of the biggest, most important and generally interesting news coming out of the first day of the show. There was a strong showing from the biggest names in the PC space, with Intel showing off its latest crop of Arrow Lake chips.

Devindra: We are here what is this, the beginning of night one of CES officially? Devindra: guess we have already suffered through basically day minus one. Devindra: One thing I want our listeners to understand is that we have already seen a lot of things we kind of know where the CES is headed. And, I think this is a cursed show Cherlynn. How do you feel about that? Yeah, I think I mean, Devindra, I'll let you speak to your situation, but we've had team members who have fallen deathly ill. We have also, like, people who have completely had to miss their flights, international flights. It's been quite Engadget team, but we have a really, really good team of people. Everyone's got great attitudes and, like, our spirits are high. You want to just get the stuff going.

CES 2025 has begun, which means a whole fleet of new gadgets has been unleashed onto the world. As usual, team Engadget has battled jet lag, sleep deprivation and the static shocks of those horrible casino carpets to bring you all the news that's fit to print. But if you're too busy to keep your browser locked on the site (or our handy dandy liveblog) then here's a recap. This may not be everything we covered, but it's a rundown of the biggest, most important and generally interesting news for your delectation. There was a strong showing from the biggest names in the PC space, with Intel showing off its latest crop of Arrow Lake chips.

Dell is introducing three new monitors at CES 2025, but the most interesting one includes a feature you might have already experienced with a pair of AirPods: spatial audio. The Dell 32 Plus 4K QD-OLED Monitor is the first monitor with "AI-enhanced 3D Spatial Audio," according to Dell. The monitor's 5x5W speakers are able to pull off their audio tricks thanks to "AI-driven head-tracking" that follows your head and adjust audio playback so you're always immersed. It's a system that seems similar to Razer's Leviathan V2 Pro PC soundbar from 2023, which also used head-tracking to simulate a spatial audio experience. Dell's just managed to cram the necessary sensors and speakers into a monitor, with likely tinnier sound as a tradeoff.

This paper illustrates the architecture and training of Unipa-GPT, a chatbot relying on a Large Language Model, developed for assisting students in choosing a bachelor/master degree course at the University of Palermo. Unipa-GPT relies on gpt-3.5-turbo, it was presented in the context of the European Researchers' Night (SHARPER night). In our experiments we adopted both the Retrieval Augmented Generation (RAG) approach and fine-tuning to develop the system. The whole architecture of Unipa-GPT is presented, both the RAG and the fine-tuned systems are compared, and a brief discussion on their performance is reported. Further comparison with other Large Language Models and the experimental results during the SHARPER night are illustrated.

It's the dawn of a new era in Microsoft's eyes as the first wave of Copilot PCs are now available as of June 18. This "new class of Windows PCs," as Microsoft CEO Satya Nadella describes them, contains hardware designed to run as many generative AI processes locally as possible, rather than having to rely on data centers. These laptops all have an Arm-based Snapdragon chipset from Qualcomm that includes a neural processing unit (NPU) to handle such tasks. Microsoft has stipulated that Copilot PCs need to have at least 40 TOPs (tera operations per second) of NPU performance, and 16GB of RAM and 256GB of storage at minimum. The systems each have a dedicated Copilot button on the keyboard and they're all slated to have strong battery life.

We knew more computers were coming that would feature a native version of Microsoft's AI Copilot toolset, but we didn't quite know how many were set to be announced. Companies like Dell, Acer and HP have all just announced computers that have adopted Microsoft's AI software and NVIDIA's AI hardware. The age of the AI PC is upon us. These Copilot PCs could really change how we interact with computers, bringing natural language nuance to a bevy of everyday tasks. The PCs all feature a dedicated key to bring up Copilot and are stuffed with Arm-based Snapdragon X CPUs, which include powerful onboard neural processing units (NPUs) for AI tasks.