RMDM

RMDM是一个专为越南语证据验证设计的多标签数据集，由越南国立信息学研究所开发。该数据集包含1556个样本，每个样本有两个字段：文本和标签，其中标签分为真实、误导、虚假和恶意四类。数据集通过模拟不同类型的假新闻，旨在评估大型语言模型在法律背景下验证电子信息的能力。创建过程中，从越南权威网站提取真实新闻片段，并对其进行修改以生成误导、虚假和恶意信息。RMDM数据集的应用领域主要集中在法律领域，特别是在电子证据的验证中，以提高AI模型在法律应用中的可靠性和准确性。

RMDM is a multi-label dataset specifically designed for Vietnamese evidence verification, developed by the Vietnam National Institute of Informatics. It contains 1556 samples, each with two fields: text and label, with the labels falling into four categories: genuine, misleading, false, and malicious. The dataset simulates various types of fake news, aiming to evaluate the capability of Large Language Models (LLMs) to verify electronic information within a legal context. During its creation, genuine news segments were extracted from authoritative Vietnamese websites and subsequently modified to generate misleading, false, and malicious content. The RMDM dataset is primarily applied in the legal domain, specifically in electronic evidence verification to enhance the reliability and accuracy of AI models for legal applications.