While I was building Letter Library I created a data journalism piece based on letters from the Founding Fathers. Although Letter Library isn’t around anymore, I feel the research and visualizations related to that project are generally interesting and valuable. I don’t work in academia, but I’ve never seen this kind of analysis related to the founding fathers presented at this scale.
The project is hosted on this site here.
This project took a considerable amount of time as letters are not a structured data set. There was a fair amount of cleaning and categorization that needed to be done to get the project to a useful state.
Knowing how much work went into it, I wonder how easy it may have been with new LLM technology. I suspect using GPT-3.5 would have cut down the time for this project a lot.