UNIDECOR

UNIDECOR是由斯图加特大学机器语言处理研究所创建的统一欺骗语料库，整合了来自社交媒体评论、法庭证词、特定话题意见陈述及在线策略游戏中的欺骗对话等多个领域的数据。该数据集包含164,085条记录，旨在通过分析不同数据集间的语言线索相关性，理解欺骗的差异性，并进行跨语料库建模实验，以推动欺骗检测技术的发展。数据集的创建过程涉及多种数据收集策略，如爬虫、众包等，应用于心理学、法医学、法律及计算语言学等领域，以解决人类在数字媒体中易受欺骗的问题。UNIDECOR is a unified deception corpus developed by the Institute for Natural Language Processing at the University of Stuttgart. It integrates data from multiple domains, including social media comments, courtroom testimonies, opinion statements on specific topics, and deceptive dialogues from online strategy games. This corpus contains 164,085 records. Its core objectives are to understand the diversity of deception by analyzing correlations of linguistic cues across different datasets, conduct cross-corpus modeling experiments, and thereby advance the development of deception detection technologies. The construction of UNIDECOR adopts multiple data collection strategies such as web crawling and crowdsourcing. It is applied in fields including psychology, forensic science, law, and computational linguistics to address the issue that humans are vulnerable to deception in digital media.

搜集汇总

背景与挑战

背景概述

UNIDECOR是一个由斯图加特大学创建的统一欺骗语料库，整合了社交媒体评论、法庭证词和在线策略游戏等多个领域的欺骗对话数据，共包含164,085条记录。该数据集旨在通过分析语言线索相关性来理解欺骗的差异性，支持跨语料库建模实验，推动欺骗检测技术的发展，并应用于心理学、法医学和计算语言学等跨学科领域。