Preslav Nakov

Exploring the challenges and strategies for moderating hate speech online.

Learn about the challenges and methods to improve the accuracy of LLMs.

This study examines how multimodal models handle false claims with text and images.

Introducing a method to assess reliability in language model outputs.

This article discusses a method for connecting attack behaviors to techniques using MITRE ATTCK.

A look at how to identify human and machine-written content.

Transforming an MCQA dataset for extractive questions in multiple languages.

A framework to ensure language models provide accurate information.

A deep dive into meme analysis and its societal effects.

This study examines methods to enhance machine empathy through storytelling.

This article discusses the importance of measuring uncertainty in AI predictions.

New dataset improves how models convert web pages into HTML code.

Researchers develop methods to improve language models for various languages.

OpenFactCheck provides a framework for evaluating the accuracy of language model outputs.

A detailed approach to identify machine-generated texts effectively.

A method to improve how confident language models are in their text generation.

This project enhances text correction in Bulgarian historical documents using OCR technology.

TART enhances table reasoning tasks using specialized tools and large language models.

Learn how social event detection works and its importance in today's world.

Multi-agent systems help robots learn and adapt while working together.

MediaGraphMind helps evaluate news source reliability and bias effectively.

RAG improves language models but faces challenges from misinformation attacks.

Researchers propose new methods to keep LLMs safe from harmful content generation.

A new framework prioritizes safety alongside performance in AI evaluation.