It seems that this might indeed be the case. After writing the comment I’ve had a look through the (obsolete) Ruby code base of the project that somebody linked

It seems that this might indeed be the case. After writing the comment I’ve had a look through the (obsolete) Ruby code base of the project that somebody linked, and it doesn’t actually perform (what I’d call) text mining as far as I can see. Instead it parses the court documents, which seems to be a mostly structured format. No natural language processing.