The corpus consists of over 3.5 million files, totaling 30.7 GiB of raw text, collected from 159 sources encompassing judicial, legislative, academic, news, and other related data.

https://doi.org/10.1007/s10579-024-09762-8