Just poking around on the internet for examples of topic modeling in history, and found some interesting ones I had never seen before. The most affecting was the analysis of the famous Martha Ballard diary by historying, Cameron Blevins. From his topic modeler, Blevins got one group of words from Ballard’s diary, “feel husband unwel warm feeble felt god great fatagud fatagued thro life time year dear rose famely bu good.” Essentially, this meant that these words appeared together frequently in the text as a “topic” that the computer could recognize again and again. And as Blevins judged, the most descriptive word he could come up with for this computer-identified topic was “EMOTION.” When he charted the topic over time, he was able to show, very clearly and interestingly, that Ballard’s diary reflected more emotion over time. It’s a very interesting analysis. Certainly not definitive, but intuitively very convincing. I found it powerful.