Exploring automatic suggestions

One of the projects I’m excited about is an experiment with providing automatic suggestions to student editors about how to get started with improving their assigned Wikipedia articles. We want the dashboard to highlight specific improvements that student editors can make to their assigned articles, such as adding additional sources, including an image.

I’ve been mentoring a new contributor to the Dashboard codebase, Keerthana S, to build out a beta version of this feature, and we’re currently testing it out with a few of our summer courses. Our hope is that this feature will help Wikipedia newcomers get started more quickly, focus their efforts on the most impactful improvements. and ultimately spend more time on the substance of their topic, and less time mastering Wikipedia-specific conventions and style issues.

Keerthana, a sophomore Engineering Physics student at the Indian Institute of Technology, Madras, is spending the summer working on the Wiki Education Dashboard as a Google Summer of Code intern. So far, she’s built a way to display article suggestions to students, based on the current state of each of their assigned articles. It provides basic suggestions based on Wikimedia Foundation’s artificial intelligence article scoring platform, ORES. The early suggestions are very rudimentary, focused on easily detectable problems like when an article lacks references or ought to be broken up into sections. Now that the framework is in place, Keerthana will be working on making the suggestions smarter, more diverse, and more nuanced.

We’d love feedback from users at this point: what kinds of suggestions will be most useful to students? What patterns in student work could we detect? What problems could this feature head off? You can browse any current or previous courses, go to the Articles tab, and then look at the ‘Assigned Articles’ section to see the suggestions. (For example, this summer course about ancient Rome.) There’s a form to submit your feedback right from there, and any ideas you have will be helpful — especially for any classes you’ve participated in and the articles your class worked on.

Keerthana is also planning to add the ability for instructors and other students to provide their own suggestions from within the Dashboard on how to improve specific articles.

If you want to dive into the nitty-gritty details of Keerthana’s project, you can follow her blog, where she keeps a weekly journal of her work and what she’s learning.


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.