Eliana Gelman, Franklin College of Arts and Sciences, Institute of Artificial Intelligence, Cognitive science, 4th year undergraduate
When speaking to my mentors and peers, I realized that there was often a lack of communication across departments, even amongst people investigating the same field from different angles. As someone who was always interested in interdisciplinary work, I felt there was room to improve here. I made the UGA Scholar Network as an interactive web application that visualizes research connections between UGA faculty members across departmental boundaries. Instead of clicking on individual profiles alphabetically in a directory, this app transforms complex faculty research descriptions into an intuitive 3D visualization where proximity indicates research connections.
The system works by collecting faculty information through web scraping from various UGA department websites, then using Google’s Gemini 1.5 Flash AI to analyze and summarize lengthy research descriptions. These summaries are converted into vector embeddings using TensorFlow.js and the Universal Sentence Encoder, with similarity between researchers calculated using cosine similarity. Finally, these relationships are projected into a 3D space using Three.js.
In the visualization, each faculty member appears as a glowing node with size representing research prominence and color indicating research domain. Users can hover over nodes for basic information, click for detailed profiles, and search for specific faculty or research interests.
The platform serves multiple audiences: students can find potential advisors across departments based on research interests; faculty can discover collaborators with complementary expertise; administrators can identify strategic areas for interdisciplinary initiatives.
Technical features include AI-powered research summarization and semantic analysis, interactive 3D visualization with intuitive navigation, responsive design for various devices, and accessibility features for inclusive use. People can also search for research topics and find the matching nodes as a result. I hope to expand this network to populate it with data on all faculty research at the University. This current version is focused on the Institute of AI. The UGA Scholar Network will help more innovative and collaborative research take place by revealing intellectual connections that departmental divides and information overload often obscure. My ultimate goal is to create an interface that increases collaborations and strengthens UGA’s research ecosystem.