To start this job, select the question: “Given two sets of articles, what words mark out an article as belonging to set A rather than to set B, or vice versa?”
The algorithm implemented here is Zeta, originally described by John F. Burrows in 2007 and expanded by Hugh Craig, as implemented by David L. Hoover. This algorithm takes two datasets as input (call them A and B), and returns two lists of words. Each list of words is a set of words – which are neither particularly common nor exceedingly rare – that are likely to mark a text out as belonging to set A or set B respectively. That is, if “Germany” is a Zeta word for set A, then the appearance of “Germany” in a text makes it much more likely to belong to set A than to set B.
This algorithm can be used to answer the following kinds of questions:
What terms are commonly used within one set, but rarely used outside of it? (Input: two datsets, one set of interest and one set containing the rest of the corpus)
What terms make one strand of discourse different from another? (Input: two datsets, one each from each strand of discourse)
What concepts have entered or left a discipline over time? (Input: one dataset of earlier works, one dataset of later works)
This analysis has no options.