I liked reading the (unnamed?) paper by Warren et al.\u00a0 The main goal of their project was to digitally create a map of how early modern people were related and interacted.\u00a0 They describe a number of challenges they face in doing this, such as the lack of documented info from some time periods.\u00a0 Although I couldn’t fully understand all of their statistical techniques, I understood their methodology for the most part.\u00a0 One important takeaway was not inferring causation from correlation.\u00a0 In their example, they show how two people could be mentioned by the same person, but could have no ties to each other.\u00a0 They also used topic modeling which we have recently been talking about and working with.<\/p>\n\n

nickbeati talked about “margin of error” in their modeling.\u00a0 They deemed a 7% margin of error from looking at a random 200-entity set and reviewing for accuracy.<\/p>\n\n

I enjoyed reading this because I had never thought of this idea as a research topic.\u00a0 I imagine doing this project in a hundred years for our time period will be much easier!<\/p>\n\n