Over the course of the Second World War approximately 127.2 million people were mobilized. The world's population in 1940 was roughly 2.3 billion, meaning that between five and six percent of the world was drafted into the military in some capacity. Approximately one in every 25 people mobilized were women, who generally served in an administrative or medical role, although hundreds of thousands of women did see active combat. Largest armies In absolute numbers, the Soviet Union mobilized the largest number of people at just under 34.5 million, and this included roughly 35 percent of the USSR's male population. By the war's end, more Soviets were mobilized than all European Axis powers combined. However, in relative terms, it was Germany who mobilized the largest share of its male population, with approximately 42 percent of men serving. The USSR was forced to find a balance between reinforcing its frontlines and maintaining agricultural and military production to supply its army (in addition to those in annexed territory after 1941), whereas a large share of soldiers taken from the German workforce were replaced by workers drafted or forcibly taken from other countries (including concentration camp prisoners and PoWs). Studying the figures The figures given in these statistics are a very simplified and rounded overview - in reality, there were many nuances in the number of people who were effectively mobilized for each country, their roles, and their status as auxiliary, collaborative, or resistance forces. The British Empire is the only power where distinctions are made between the metropole and its colonies or territories, whereas breakdowns of those who fought in other parts of Asia or Africa remains unclear. Additionally, when comparing this data with total fatalities, it is important to account for the civilian death toll, i.e. those who were not mobilized.

This dataset contains code and metadata for the paper "Vectors of Violence: Legitimation and Distribution of State Power in the People’s Liberation Army Daily (Jiefangjun Bao)." We provide the MALLET topic model, the lists of Chinese stopwords and violent terms, the Python code for acquiring the journal articles, the Python code to reproduce the results analyzed in the paper (two tables and three figures), the fine-tuned BERT model, and the metadata for all PLA Daily articles considered in the article, together with their respective violence scores.