How to use whoVIS

The "Interaction Graph" view.

The "Word Ownership" view.

The "Additional Metrics" view.

  • Explanation of metrics.
  • Length change: Gives the percentage of article growth or shrinkage in byte length, compared to the previous revision.
  • Edit rapidness: Measures the time in hours that has elapsed since the previous revision.
  • Average editors disagreed with (window=50): Shows how many editors' content was affected on average per revision by the actions in the last 50 revisions. (i.e., "how many 'feet were stepped on', on average, in the last 50 revisions")
  • Gini outgoing disagreement actions: For previous the 50 revisions, based on the interaction network, shows how equally distributed the outgoing disagreement edges are over the editors. I.e., if most or all of the negative actions came from just a few editors, the Gini coefficient will converge to →1. If the disagreement actions came equally from all editors, it will converge to →0.
  • Gini incoming disagreement actions:For the previous 50 revisions, based on the interaction network, shows how equally distributed the incoming disagreement edges are over the editors. I.e., if most or all of the negative actions target just a few editors, the Gini coefficient will converge to →1. If the disagreement actions equally target all editors, it will converge to →0.
  • Gini editorship: Taking as a base all unique editors ever active in the article, this measure shows how equally all edits made in the article are distributed over the editors. I.e., if most or all of the edits were made just by a few editors, the Gini coefficient will converge to →1. If the edits were made at equal proportions by all editors, it will converge to →0.
  • Gini editorship (window=50): Limits the Gini editorship to only the previous 50 revisions and reports the average.
  • Template:**** curves: Show for several important templates when they were present (value=1) or not present (value=0) in the article (or its talk page, if template doesn't appear on main page)

Best practices and notes.

  • Drag nodes to explore crowded clusters. As "neutral" nodes in the network graph are placed in the center of the drawing area, the latter naturally tends to get crowded. You can drag nodes away from their original position to get a better overview. Note, however, that the relative positions of moved nodes to all other nodes lose the inherent semantics bestowed by the drawing method. To reset the positions, the page needs to be reloaded.
  • Open different view tabs in different browser windows for comparison. Because of data load (cf. next bullet), the different data views were not placed on the same page. It is however useful (if your machine's resources allow it) to open the word ownership and/or additional metrics in browser windows underneath the interaction graph to compare the status at given revisions.
  • Large articles entail heavy loading and rendering. Displaying the network graph, edge context and times series requires a lot of data to be loaded and rendered by D3, especially in the cases of large articles/many edges/time series with many data points, so please give the tool some time in these instances. Try not to display too many of the time series in the line charts at once (toggle buttons). If you are on a Mac, Safari seems to cope best with the data load, else, please use Chrome.