This model was produced using the “Data for Research” OCR files generously provided by JSTOR. It includes only those articles designated as book reviews, and the text has been processed to exclude common words (e.g., “the”, “Jews”) that provide less informative models. This is an interactive visualization produced through an LDA algorithm and pyLDA vis (see more on the Topic Model page for what this means). Most of these topics make sense as topics, but some do not. The computer, of course, does not know the difference between what we would call a coherent topic as opposed to the simple correlation and frequency of words.
The model “maps” the topics by the overlap between the words and their frequencies in the articles that have been sorted into that topic. Clicking on topics (which can also be cycled through with the button near the top) will bring up the words found in that topic, which in turn can be explored further. The size of the circle indicates the prevalence of the topic in the corpus being considered. One can also adjust the lambda parameter in the upper right. This parameter reorders the words to reflect the ratio of their frequency in the topic articles as compared to their frequency in the entire corpus. A fuller explanation of LDAvis visualizations can be found here.
What does it all mean? We encourage you to play around, come up with your own hypotheses, and add them to the “Participate” page!
