Behind the Scenes of Semantic Scholar’s New Author Influence Design
By Cassidy Trier, Senior Product Designer • Evie Cheng, Senior UX Researcher • Ashley Lee, Marketing Specialist
Semantic Scholar’s author pages are a source of unique information about authors and the impact of their work. Our team continues to develop new features to help scholars better understand their own work and the work of their colleagues.
Earlier this year, we released a new version of our Author Influence visualization based on user research insights.
Problem
At Semantic Scholar, we listen to scholars’ needs to improve our research tools. Our author influence visualization was originally developed as a way to show aggregate scores which calculate the authors who most influence and are most influenced by a particular author. However, many users were confused by what influence means and how it’s measured.
“My primary issue with the influence. I just have a lot of questions about what that specifically means.” –User study participant
Design Process
We turned to user research to learn how scholars construct mental models of authorship networks. In the first iteration, we decided to replace a composite influence score with more specific relationship types to make the nature of the “influence” easier to understand, and highlighted four kinds of author relationships:
- Citing Authors
- Referenced Authors
- Mentorship
- Co-Authors
We explored different layouts that visualize the influence information, one was an improved network graph visualization, and the other was a list view.
Based on user feedback, we decided to prioritize the list view because it was easier for users to navigate, more accessible for people who use tabbing navigation or screen readers, and also compatible with small screen sizes, such as mobile phones.
We also removed the mentorship data because we learned that this kind of data is sensitive and errors (e.g., labeling a colleague as a mentee) could lead to negative reactions.
Additionally, some people had difficulty discovering this feature on author pages. To remedy this, we created individual tabs for each author relationship which act as both a tool for navigation between pages and also as a preview of the utility of the feature.
Next Steps
Although we haven’t seen a significant increase in engagement metrics since the launch, we have observed qualitatively added value of the feature to our users.
In a recent user survey, the author influence list was rated as one of the top 5 most valuable features on Semantic Scholar. About 24% of our users said author influence lists are valuable to them. Compared to the results of 15% from a year ago, this is a signal that we’re headed in the right direction.
We hope you enjoy the new author influence lists. We plan to continue to optimize and iterate using A/B testing techniques. Because of this, the designs you see in this post may look slightly different from what you see on Semantic Scholar.
This is only the start of new developments on our author pages. We have several new features in the works, and we can’t wait to share them with you when they are ready.
Got an idea to make author pages even better?
Join our Beta Program
Participate in our studies and try new Semantic Scholar features before they release! Join today.
Follow @allen_ai and @semanticscholar on Twitter/X, and subscribe to the AI2 Newsletter to stay current on news and research coming out of AI2.