The User Experience of Recommender Systems

We are designing methods to understand the usefulness and usability of algorithmically generated reading and editing recommendations.

recommender systems UX image

Project overview

Wikipedia contains more than forty million articles across nearly three hundred languages. New articles are being added every day and there are still many subjects that need better coverage. Wikipedia users and contributors often find themselves asking, "What should I read or edit next?"

The Wikimedia Foundation is building algorithmic models that recommend articles for people to read, edit, translate, and create based on their personal interests. The goal is to help users filter out the information they are not interested in, so they can discover content that is relevant to them. There are many ways to generate recommendations and many ways to share them with users.

We are working to discover better ways to generate and present recommendations for Wikimedia content. We are developing new methods to predict and evaluate how recommendation-based features will be received and used by different types of users. We are also working on ways to detect and correct for potential biases in our algorithmic models, which could impact the types of recommendations users receive.

Recent updates

  1. Battle of the feeds

    We tested whether mobile app users preferred a "top article" feed ranked by page views or editing trends. Overall, raters were more interested in, and more familiar with, articles that appeared in the "top read" list.

Project team

Jonathan Morgan


Ellery Wulczyn (Wikimedia Foundation)

Resources and links