Structured Parallel Coordinates: Parallel Tag Clouds

© Copyright 2010-2011 Accademia Europea Bolzano

Structured Parallel Coordinates home page | EURAC LInfoVis page

The purpose of this Structured Parallel Coordinates visualization is to compare frequencies using the Parallel Tag Clouds model (Collins et al. 2009). (See also our RankComparison.) Click to see the modal example. (Data from the UKWAC100M web corpus of British English.)

Click to show instructions

Click to hide instructions

Each set of data is entered in the areas below line by line, in ranked order, from first to last, where each line is in one of these three formats:

  1. numerical value followed by a tab followed by items with that numerical value (same rank), with tabs between them; the numbers can indicate, i.e. frequencies, statistical measures, wave length, size in cm, etc.
  2. a single item followed by a tab followed by a number
  3. items that have the same rank separated by tabs, with no number

As the tab key cannot be used when entering data, it is best to prepare the data in a text editor and copy it to the data field.

"Number first" should be checked for the column in the first case, and it should not be checked in the other two cases. Items with the same number will automatically be treated as having the same rank (as long as they are entered on subsequent lines). Blank lines are ignored.

You can also provide a label for the data, as well as choose whether the data will be included in the visualization. You can change the order of the data sets by dragging the column headers ("Data 1", etc.) to the desired place.

Once you have entered your data, click on "Compare ranks" to see the comparison below the data chart. The items are listed in alphabetical order and scaled by their relative frequency, either "by all" results, or "by dimension". In addition, the scaling can be linear (by raw frequency) or logarithmic (to take into account Zipf's Law). Identical items in different data sets are connected by lines. ("[NA]" means that the relevant item does not occur in the series of that axis.)

Data 1Data 2
Number first

Scale by all Scale by dimension || Log scale Linear scale