Tracking the Narrative: A Data-Driven Analysis of Media Coverage of Russia and Ukraine 2013-2024

This dataset contains over 22 million news articles mentioning Ukraine and Russia, published between January 2013 and December 2024. It served as the primary input for the publication 'Tracking the Narrative: A Data-Driven Analysis of Media Coverage of Russia and Ukraine 2013-2024.' The data are organized by month, and for each article, the following information is provided: Title, Link, Publication Date, Country, Language, and Cluster. Articles with publication dates after July 2019 also include a list of entities mentioned in the text. By tracking media coverage over time and employing multilingual clustering along with large language model (LLM) summarization of clusters, analysts identified key geopolitical events reported in the media.

The 'cluster' field provides information derived from the clustering algorithm applied to the article titles. Each cluster can be considered a story or narrative that was prevalent during the given month. The LLM analysis for each month is saved in separate files containing the following fields:
'Cluster' (serves as an ID to connect with the main data file),
'Cluster Title' (LLM-generated title for the cluster/story),
'Cluster Title Short' (LLM-generated short title for the cluster/story, used in visuals),
'Cluster Keyphrases' (Five most relevant key phrases for the cluster/story generated by LLM),
'Count' (Total number of articles in the cluster).

The dataset also contains HTML files for the visuals used in the publication. Interactive visuals facilitate the exploration of the study's results, providing a comprehensive analysis of media narratives related to Russia and Ukraine from 2013 to 2024. The visuals illustrate the distribution of articles over time and highlight the most prevalent narratives for each month. Furthermore, they display the results of custom cluster-linking techniques, visualizing the evolution of stories over time.