Researchers have introduced MBFC-2025, a large dataset of approximately 2,600 media outlets, to address the lack of unified resources for detecting political bias and factuality in news. The study also constructs multi-view representations for existing datasets like ACL-2020 and MBFC-2025, incorporating various data sources such as hyperlink graphs and LLM-derived information. A systematic evaluation of embedding views and fusion strategies was performed, achieving state-of-the-art results on ACL-2020 and setting new benchmarks for MBFC-2025. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Provides new resources and benchmarks for developing AI systems capable of identifying political bias and factuality in news media.
RANK_REASON The cluster contains a new academic paper detailing a dataset and evaluation methodology for media bias detection. [lever_c_demoted from research: ic=1 ai=1.0]