This site is an automatically generated index of themes in a collection of more than 130,000 research articles archived in JStor. For each theme, you can browse articles associated with that theme, with links directly to the text of the articles in JStor. You can also view temporal trends in publication for the theme, organized by journal.
Each theme is represented by assigning a weight to every word in the vocabulary of the corpus. The 50 highest weighted words are shown. Some themes are better-estimated than others. Words that are shown in full-size and black are likely to be meaningful, while smaller words in lighter shades may be random artifacts of the model.
The time-series plot shows the proportion of the words in the corpus in a given year that are assigned to a theme. Gray bars show decades. The y-axis is not comparable between themes.
The work presented here was developed by David Mimno under the Cybereditions Project, an effort led by the Perseus Project at Tufts University and funded by the Mellon Foundation.