Editor Growth and Contribution Program/Taghreedat

From Meta, a Wikimedia project coordination wiki

Taghreedat - Wikipedia Editors project is a pilot project which is being executed in collaboration between Taghreedat initiative and the Wikimedia Foundation. The purpose of this project is to attract and train new editors on how to edit Wikipedia, where Arabic Wikipedia is being used as a platform to run the current pilot project. The idea of the current pilot project is to attract new users to Wikipedia using Taghreedat’s Twitter account, and to provide them with a step-by-step training on how to edit Wikipedia by sending them specific tasks by email. The pilot project which is documented here was run over a period of 4 months between 1st June and 30th September 2012.

This report is an interim report as the project is still running, and aims in sharing some of the key learnings that include:

  • Social media channels, such as Twitter, could be very powerful in attracting new enthusiastic editors to Wikipedia.
  • Sending emails to new editors with specific tasks along with tutorials, resulted in a high response rate on editing Wikipedia pages.
  • Coaching new editors on various Wikipedia editing techniques using a step-by-step approach resulted in converting a considerable portion of them into repeated editors.
  • A greater percentage of editors from this program edit more often than an average new user. This might be an early indicator that a project like Taghreedat - Wikipedia Editors project could make a meaningful contribution to the community of active Wikipedians.

Background[edit]

The original proposal for Taghreedat and Wikipedia Cooperation.

Taghreedat, a non-profit Arabic digital content community building initiative, has launched the Arabic Wikipedia Editors Program which targets regular Internet users and train them (both online & offline) to become editors on Arabic Wikipedia. Taghreedat’s Twitter account has currently more than 100K followers. Taghreedat initiative is aiming to form a task force from among these volunteers for the purpose to promote the Arabic contents on the Internet in various fields.

Key idea[edit]

The key idea of this pilot project is to attract new editors to Arabic Wikipedia from among the followers of Taghreedat’s Twitter account, provide them with the proper training via online means (e.g. mails, tweets) and guide them through the basic steps of editing Wikipedia by requesting them to perform simple tasks on Wikipedia.

Workflow[edit]

Workflow of Taghreedat-Wikipedia editor project.

The workflow of Taghreedat Wikipedia editor project consists largely of two steps, the registration step, and the learning pipeline, and is illustrated roughly in the figure on the right.

Registration[edit]

The registration step started by posting tweets on Taghreedat’s Twitter account asking its followers to fill a registration form in case they are interested in contributing to a project for enriching the contents of Arabic Wikipedia. The tweet is scheduled to be re-posted frequently in order to attract attention from a large number of Taghreedat’s Twitter followers. The registration form can be found at this page.

During the registration process, users were asked simple questions about their names, twitter account, email address, country of residence, and input their Wikipedia user name after registering an account. These information were collected and maintained by Taghreedat team, and they shared only a list of usernames with the Wikimedia Foundation for the purpose of data analysis.

Learning Pipeline[edit]

After completing the registration process, data provided by the contributors was stored into a database which was used to contact them during the learning pipeline in order to coach them on how to edit Wikipedia. The learning pipeline consists of multiple steps starting from easy tasks which get harder by time.

Each task is sent to the project participants via emails. Each email included a description of the task, description of its importance, a tutorial or a help page to assist them perform the task.

Completed tasks[edit]

  • Create a user page
  • Add a wikilink to a dead-end article
  • Add a category to an uncategorized article

Proposed Tasks[edit]

Following are some of the proposed tasks for later phases of the project. The sequence is arbitrary and might be altered while monitoring the response of the participants.

  • Improve an article stub
  • Create a page on Wikiquote
  • Upload a photo taken at their own city
  • Add links to an orphaned article
  • Add languages links to an article
  • Copyedit an article
  • Add a reference
  • Translate an article
  • Post a message on a other user’s discussion page

Progress[edit]

First task : create a user page[edit]

The first task was creating a user page. An email was sent on July 7th to about 1000 contributors who registered their email addresses during the registration phase. The email included a short text asking each user to create their user page on Arabic Wikipedia, and provided a link to a step-by-step page tutorial on how to create a user page. The email was made brief and informative in order to direct users to the tutorial page and let them learn how to create their user page by themselves. Since creating a user page doesn’t require contributing to the article namespace, it is one of the ideal tasks to start coaching new users on. We noticed that a large number of users managed to create their userpages, added userboxes to it, and wrote a brief introduction about themselves and their motivations on editing Wikipedia. Below are a few quotes from the created userpages :

  • I am delighted to join this campaign to boost the Arabic content of this great encyclopedia (User:Miyadeh)
  • I feel it is my responsibility to enrich Wikipedia with more contents in science, thus more people will be able to benefit from it (User:ahmad.aea.99)

Second task : Add internal links to articles[edit]

Wikipedia receives a large number of new articles which are acceptable by their contents but contain no internal links to other Wikipedia articles (aka: wikilinks). Such articles are known as dead-end pages. These pages require someone to add links to them. We designed the second task to provide each volunteer with a dead-end page and ask them to add internal links to them. Similar to the first task, an email was sent to all volunteers on Sep, 3rd including a tutorial page on how to add a wikilink to a page and a link to a randomly generated page from a maintenance backlog category containing all dead-end pages using the random article tool.

Third task : add categories to articles[edit]

Similar to dead-end articles, Wikipedia receives a large number of new articles which lack any category. In a similar manner to the second task, we designed the third task to provide each volunteer with an article that has no categories and ask them to add one or more relevant categories to it. An email was sent to all volunteers on Sep, 29th including a tutorial page on how to add a category to a page and a link to a randomly generated page from a maintenance backlog category containing all uncategorized pages using the random article tool.

Data analysis[edit]

Edits[edit]

Figure 1 : Total number of edits made by Taghreedat volunteers on Arabic Wikipedia, along with the daily average of edits in each stage of the program.

Until October 2nd, there were 745 user accounts registered in Taghreedat program (The full list of user-names can be found at this page), contributed a total of 3410 edits on the Arabic Wikipedia so far. 50 users out of the total users contributed to Arabic Wikipedia a total of 629 edits before the program started (18% of the total edits), which means that about 695 new user accounts and 2781 edits were contributed thanks to Taghreedat program.

Figure 1 shows a plot of the total number of edits by date which increases from 695 edits at the time before starting the program, up to 3410 edits by the time of preparing this report. The major milestones which contributed in the increase in number of edits can be noticed when users started registration, and further increase when users responded to the tasks of creating their userpages, wikifying articles, and categorizing them. Figure 1 also shows the average number of daily edits which resulted in a remarkable increase from 7.8 edits/day during the registration stage up to 36.3 edits/day at later stages of the program, which shows that more users have joined the program since it’s started, and they are doing more edits as they are more familiar with Wikipedia editing skills.

Figure 2 : Total number of edits made by Taghreedat volunteers on daily basis.

Similarly, Figure 2 shows the total number of edits by day made by all contributors. Peaks in the graph reflects the instant response of a large portion of contributors to tasks sent to them by email, we can notice that the registration stage, which was promoted mainly by tweets on Taghreedat’s Twitter account, has motivated interested users to register their accounts and initiate their first edits without having a definite clear task. In contrast, the same figure shows that asking users directly to do certain tasks on Wikipedia drove more number of edits compared to the stage of promoting Wikipedia on a social media channels such as Twitter.

Editors[edit]

Figure 3 : Proportion of Taghreedat users by the number of their edits on Arabic Wikipedia during the program.

As mentioned earlier, the total number of registered users during this pilot in Taghreedat program was 745. By analyzing the number of edits of these editors, we can notice that about 43% of these registered accounts didn’t do any edits during the program so far. On the other hand, about 20% of the participants made more than 4 edits. Assuming that each task requires at least one edit, we can conclude that a large portion of the users made more edits than the required number of edits to perform the task. However, by observing contributions of a few number of users, we noticed that the first and second the tasks (Creating a user page, and adding wikilinks) required the user to make more than one edit in order to get a satisfactory result on the page.

Figure 4 : Comparison between emailed users and average users by their number of edits.

In order to understand the significance of these results, we compared participants of this program with the sum of all users of Arabic Wikipedia categorized by their number of edits. This comparison is presented in Figure 4. We can notice that while only 1.7% of all users made more than 10 edits, 6.3% of the participants in this program made more than the same number of edits. In contrast, while 89.8% of all registered accounts on Arabic Wikipedia have never made any contribution, only 42.6% of the participants in this program didn`t make any edits.

Although the percentage of participants who made no edits is about 43%, we believe that many users did actually make some edits in response to tasks sent to them by email. However, we assume that they didn`t log in their accounts at that time, which make this percentage lower in the actual case.


Figure 5 : Proportion of Taghreedat users by the number of their editing days on Arabic Wikipedia during the program.

Figure 5 shows the participants categorized by the number of days they edited Wikipedia. Categorizing users by the number of their editing days is more accurate compared to categorization by the number of edits, because it eliminates any bias to repeated editing in the same page. Assuming that the participants responded to each required task by making edits in a single day (the day which they received the email, read it, and edited Wikipedia) we can notice that 6.7% of the participants responded to the three tasks sent to them so far, and 6.6% of the total participants edited Wikipedia in more days than the number of tasks. We assume this category of users edited Wikipedia in a number of days more than the number required by the tasks, have started to become repeat Wikipedia editors.


Media Coverage[edit]

Next steps[edit]

As the number of registered users increases, it might become difficult to coach, communicate, and collect data using the manual methods which we are utilizing in the current pilot. We are aiming at building an automatic system that offers a step-by-step training for interested users, while monitoring their performance, and collecting feedback from them. As for attracting more registration in this project, we are planning to drive more users from the Contribution portal which is a landing page linked from the site banner. Other potential pages for driving registration may include the community portal, and similarly functioning other Wikipedia projects pages.