Authors: Qiang Liu, Xiang Tao, Junfei Wu, Shu Wu, Liang Wang

Published on: February 06, 2024

Impact Score: 8.22

Arxiv code: Arxiv:2402.03916

Summary

• What is new: Introduces an LLM-empowered approach for rumor detection on social media that leverages prompts and a Chain-of-Propagation strategy to improve detection performance.
• Why this is important: Difficulties in enabling LLMs to sift through complex social media propagation information for rumor detection, due to the overwhelming amount of news contents and comments.
• What the research proposes: LeRuD, an approach that utilizes designed prompts to guide LLMs in focusing on crucial clues and divides the information into a manageable Chain-of-Propagation.
• Results: Achieved improved performance over existing state-of-the-art models by 2.4% to 7.6% on the Twitter and Weibo datasets, showcasing enhanced rumor detection in few-shot or zero-shot scenarios without needing any data for training.

Technical Details

Technological frameworks used: LLM-empowered Rumor Detection (LeRuD)

Models used: Large Language Models (LLMs)

Data used: Twitter and Weibo datasets

Potential Impact

Social media platforms, News organizations, Online community management services, Fact-checking services

Want to implement this idea in a business?

We have generated a startup concept here: RumorGuard.