BestRef

📢 #BestRef July 2026 update! 🔬4.2 million websites from more than 400 million #Wikipedia references across 300+ language editions were analyzed. 🌍 This update provides a unique view of the sources @Wikipedia relies on most. 🔍 Try it now: bestref.net

📰🔍 Accuracy and #PoliticalBias of News Source #Credibility Ratings by Large Language Models. 🤖⚖️ 📊 Left: correlation among different #LLMs and human experts. 📈 Right: ratings generated by LLMs with different partisan roles. Source: dl.acm.org/doi/full/10.1145/…

#Polarization and #reliability of news sources in #Wikipedia: top 5 news outlets reliability class for different political polarization groups, including @nytimes, @Telegraph, @washingtonpost, @timesofindia, @FoxNews. Source: doi.org/10.1108/OIR-02-2023-…

Finnish #Wikipedia - information about 3.37 million references was extracted from 559.93 thousand articles. This allowed to identify and assess 138,320 websites as sources of information. bestref.net/fi

Serbian #Wikipedia - information about 3.58 million references was extracted from 674.48 thousand articles. This allowed to identify and assess 102,246 websites as sources of information. bestref.net/sr

Vietnamese #Wikipedia - information about 3.8 million references was extracted from 1.29 million articles. This allowed to identify and assess 147,041 websites as sources of information. bestref.net/vi

Catalan Wikipedia - information about 3.9 million references was extracted from 735.55 thousand articles. This allowed to identify and assess 197,470 websites as sources of information. bestref.net/ca

Ukrainian #Wikipedia - information about 5.46 million references was extracted from 1.29 million articles. This allowed to identify and assess 217,787 websites as sources of information. bestref.net/uk #verifiability #reputation

Arabic #Wikipedia - information about 6.36 million references was extracted from 1.22 million articles. This allowed to identify and assess 294,089 websites as sources of information. bestref.net/ar

Portuguese @Wikipedia - information about 7.69 million references was extracted from 1.11 million articles. This allowed to identify and assess 319,534 websites as sources of information. bestref.net/pt #datascience #reliability #bestref