The Russian-Ukrainian war has attracted considerable global attention; however, fake news often obstructs the formation of public opinion and disseminates false information. To address this issue, we have curated the RUWA dataset, comprising over 16,500 news articles covering the pivotal events of the Russian invasion of Ukraine. These articles were sourced from established outlets in the USA, EU, Asia, Ukraine, and Russia, spanning the period from February to September 2022. The paper explores the use of semantic similarity to compare different aspects of articles from various web sources that cover the same events of the war. This unsupervised machine learning approach becomes crucial when obtaining annotated datasets is practically impossible due to the lack of real fact-checking during the ongoing war. The research goal is to uncover the potential of employing semantic similarity measures as a viable approach for detecting misinformation in news articles.
Unsupervised approach for misinformation detection in Russia-Ukraine war news
Lo Scudo F.;
2024-01-01
Abstract
The Russian-Ukrainian war has attracted considerable global attention; however, fake news often obstructs the formation of public opinion and disseminates false information. To address this issue, we have curated the RUWA dataset, comprising over 16,500 news articles covering the pivotal events of the Russian invasion of Ukraine. These articles were sourced from established outlets in the USA, EU, Asia, Ukraine, and Russia, spanning the period from February to September 2022. The paper explores the use of semantic similarity to compare different aspects of articles from various web sources that cover the same events of the war. This unsupervised machine learning approach becomes crucial when obtaining annotated datasets is practically impossible due to the lack of real fact-checking during the ongoing war. The research goal is to uncover the potential of employing semantic similarity measures as a viable approach for detecting misinformation in news articles.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
