Discipline Browser

Users can search for articles using key terms on JSTOR's main site, but many of these search results may relate to themes outside of the user's interests. The Discipline Browser is an experimental search engine which allows users to search for key words in the context of certain themes.

The themes we select are based on overall word counts for our journals' disciplines. We use topic models to determine the weight of each theme for each of JSTOR's articles.  These themes are then used to help refine a user's search results to provide more focused searches.

An article's theme weights are based on its words, not labels attached to its journal.  Therefore the themes of an article are based purely on the text of the article, not on subjective labels applied by humans.

With the discipline browser, you are presented with a search box and a list of disciplines.  You begin by entering a search term (currently only single-word queries are supported).  You are then presented with a list of themes which correspond to the search term provided.  You can adjust the weight of any of these themes, or you can adjust any theme from the list of inactive themes.  Each time you tweak the weights of these themes, the browser updates the list of search results to better fit your search needs.

This experimental index contains only a random subset of JSTOR's collections.