The program is based on the algorithms described by Jiří Milička in his article Minimal Ratio: An Exact Metric for Keywords, Collocations etc (CSLR 2012).

A) Input

  1. Choose the text(s) you want to explore. Use multiple selection if you want to add more than one text.
  2. Choose the reference text(s) (the first text will be compared with it). Use multiple selection if you want to add more than one text.
  3. Specify the type of the texts in the comboboxes. Concordance in the following formatting can be also used:
    WordType1 8
    WordType2 19
    (The word types should be delimited by a tab from their frequencies, each entry on a new line)
  4. Specify the case sensitivity.
  5. Specify the confidence level (95 % by default).
  6. Specify whether the reference corpus contains the text to explore (the second text is a submultiset of the first text).
  7. Press the "Find the keywords" button.

B) Processing

The algorithm is quite greedy so please be patient when exploring long texts (the progress is not visible).

C) Output

  1. Output data will be shown in separate window automatically after the finish of the process. You can save it and load and sort the table according to the columns in both directions.
  2. The "Show results" button shows the results when they are not visible.