Research:Editor trajectories

From Meta, a Wikimedia project coordination wiki

This page documents a research project in progress.
Information may be incomplete and change as the project progresses.
Please contact the project lead before formally citing or reusing results from this page.

Wikipedia (WP) editors contribute to WP projects as anonymous editors or by creating accounts. Little is known about the possible trajectories of editors after their account creation or first edit.

This study focuses on identifying editor trajectories and assessing their relation to an editor's involvement via edits (or lack of involvement) in project(s).

Research questions[edit]

  1. Numbered list item What are the most and least probable editor trajectories?
  2. Numbered list item Are the trajectories different across different platforms (desktop vs. mobile) and projects?


Identifying editor trajectories requires identifying the states the editors can possibly be in throughout their involvement with WP project(s). Identifying such states is not a trivial task given that there are many variables which could have impact on defining different states the editor traverses. For example, at each point of time, the trajectory of an editor may depend on the number of past edits, percentage of past edits reverted, the specific project he/she is contributing to, the gender, and more.

To find the relevant states, we first take a sample of few editors and observe their trajectory based on (edit, time of edit). In the second part, we use clustering algorithms to find the most important variables defining the states editors can be in. In this second part, we pay specific attention to editors with long term commitment to WP projects vs. those who stop editing and assess their differences in trajectories.

Sample Editor Trajectories[edit]

5 editors who have done at least one edit based on revision table are selected uniformly at random from the list of 16,535,088 editors in logging table who had created an account, i.e., log_type="newusers" and log_action="create". Note that not all the editors who create an account do edit. The randomly selected editors who never edited (13 such editors who were present in the random sampling) were discarded.