This interactive visualisation uses the text from each chapter, instead of the text from a whole document, to create clusters. This increased granularity shows that while there is some apparent split between trade agreement types, it is less clear than at the document level. Furthermore, additional data has been added to the tool tip allowing us to see the chapter titles. Doing this allows us to explore and see if certain chapter types cluster together.
The real takeaway here is the fact that once the trade agreements have been separated into chapters, it becomes difficult to clearly separate and cluster them. This is interesting as it tells us that trade agreements are in fact quite similar overall.