MBIC (Media Bias Including Characteristics)

MBIC是由德国康斯坦茨大学创建的一个包含1700条媒体偏见实例的数据集，旨在通过非专家的标注来识别和分析媒体偏见。数据集涵盖了从2019年1月至2020年6月的1000篇文章中的句子，涉及14个不同的事件和议题。创建过程中，研究团队利用自建的标注平台，通过亚马逊Mechanical Turk招募了784名位于美国的标注者，收集了详细的标注者背景信息。MBIC的应用领域主要集中在媒体偏见检测，旨在通过大规模、多样化的数据集提供更准确的偏见识别方法。MBIC is a dataset consisting of 1,700 media bias instances, developed by the University of Konstanz in Germany for identifying and analyzing media bias via non-expert annotation. It comprises sentences extracted from 1,000 articles published between January 2019 and June 2020, covering 14 distinct events and topics. During the dataset construction process, the research team employed an in-house annotation platform, recruited 784 U.S.-based annotators through Amazon Mechanical Turk, and collected comprehensive background information about these annotators. The primary application area of MBIC is media bias detection, aiming to provide more accurate bias identification methods through its large-scale and diverse dataset.

搜集汇总

背景与挑战

背景概述

MBIC是由德国康斯坦茨大学创建的媒体偏见数据集，包含1700条实例，覆盖2019年至2020年间的1000篇文章和14个事件，旨在通过非专家标注来检测媒体偏见。数据集通过亚马逊Mechanical Turk招募了784名美国标注者，收集了详细背景信息，专注于提供大规模、多样化的偏见识别方法。