Transparent, AI-powered algorithms match volunteers to citizen science opportunities

On SciStarter, personalized, recommended projects now include explanations to describe how projects are matched to logged in users.

Through SciStarter, millions of people participate in thousands of citizen science projects that engage the public to help crowdsource answers to research questions. Finding the right project for you among thousands of options can be like finding a needle in a haystack. So, for the past couple of years, SciStarter and Ben-Gurion University have tested a variety of algorithms to understand which characteristics of projects (location, topic and more) relate to activities of SciStarter users (projects they view, save, review, visit, join and participate in) to produce matches that result in a user joining and participating in a project. Learn more about that process in a past blog post.

More recently, we added conditional messaging to explain how the recommended projects are selected for a user. We will also test to see if this level of transparency and clarification results in more participation in citizen science projects.  

The image below shows the new and improved recommendations for a sample user named Daniel. Find your own recommendations on the main SciStarter website and in your own Dashboard when you make a SciStarter account.

Recommendations most often appear in batches of three projects. The recommended projects for Daniel were: Flu Near You, The Neureka Project and ZSL London Zoo. The justification for recommending these three projects is that Daniel contributed to other projects with similar features in the past that fit his interests.

Additional information can be inferred from clicking/hovering on the tooltip (the question mark next to the project name). For example, Flu Near You is all about Health & Medicine, and Daniel contributed to other health-based projects in the past. 

Another justification the algorithm uses to select projects is the link between your own interests and other users. For example, Daniel, who is fond of health-related projects, may be offered the Colony B project with the following tooltip: “People who contributed to projects involving Health & Medicine also liked Cell & Molecular projects.”

By clicking on “See more recommendations,” you can receive additional suggestions and justifications for projects, generated with your interests in mind. 

Note that you can manage or stop receiving these recommendations at any point by clicking on the relevant option. 

With the recommendation system, Daniel can also review project options that are near his location. For example, one of Daniel’s recommendations is the project Nature’s Notebook. It’s a beautiful morning in Arizona and Daniel would like to have some outdoor adventure near his home. Daniel can scroll for all his location-based recommendations and see Nature’s Notebook among his other projects under the “Projects near you” header. Then, Daniel can go outside, enjoy this project and hopefully continue turning his curiosity into impact with citizen science!

How’s your experience with the new recommendation tool? Is it helping you find projects that interest you? Email us at with your feedback! 



Categories: SciStarter News

Tags: ,

About the Author

Daniel Ben Zaken

Daniel Ben Zaken

Daniel is a MSc Student at Ben-Gurion University, Israel. His research focuses on explaining recommendation systems and he loves research and programming.