(are_s/shutterstock)
A new AI visualization tool aims to make the impact of science funding visible to policymakers, research investors and the public. Developed by researchers at Northwestern University and Tongji University in Shanghai, China, Funding the Frontier is a visual analysis tool that explores how research investments lead to outcomes across science, industry, health and policy.
According to an accompanying paper, the system links seven million grants to 140 million scientific papers, 160 million patents, 10.9 million policy documents, 800,000 clinical trials, and nearly six million news articles. In total, it maps 1.8 billion links that explore the downstream impact of research funding across multiple disciplines on how science affects society.
Many previous analyzes of science funding have focused on output within science itself, such as the number of papers or citations produced by a grant. Its developers say Frontier’s financing allows for more visibility. It measures how funded research inspires patents, informs public policy, leads to clinical advances, or reaches the media. According to the researchers, the goal is to help decision makers identify where investments generate the most benefits.
From the paper: An overview of the FTF system. The system consists of a preprocessing module, an analysis module, and a visualization module (source: paper authors).
At the core of the system is a set of algorithms that integrate and analyze data from sources including amplitude, overton, altimetric, and succinate. A large language model called Scibert, based on the Brit architecture and pre-trained on a large corpus of scientific text data, is used to read and rank millions of grant abstracts, while a machine learning method known as XGBoost predicts which projects are likely to produce high-impact results. Forecasts are displayed through a web interface that allows users to search for trends, compare funding portfolios, and identify promising researchers or topics for future support.
The system also introduces new ways to monitor the impact of grants. Each project is represented by a circular “impactglyph,” modeled on a ripple in water, showing how the grant’s impact ripples outward through publications, patents, clinical trials, and policy references. Case studies in the paper show how the system can uncover patterns such as gender disparities in funding or shifts in research focus in a field. An example shows how projects related to Alzheimer’s disease are evolving from a biomedical perspective to social and lifestyle dimensions.
Visual Design of Influence Inspired by the Metaphor of Influence (Source: Paper Authors)
The research team says it tested the system with expert users, including program officers from government agencies and executives at private science investment firms: “The system incorporates diverse impact metrics and predictive models that predict future investment opportunities into an array of integrated ideas, making it easy to search for funds and their outcomes.” “We evaluate the effectiveness and usability of the system using case studies and expert interviews. Feedback shows that our system not only meets the basic analysis needs of our target users, but also the rich datasets of complex science ecosystems and the proposed analysis framework opens new avenues for both visual and computational science.”
By linking the funding of science to its social impact, frontier funding can offer a new way to look at the flow of knowledge and resources that lead to discovery. The authors describe it as both a research framework and a practical tool for evidence-based science policy. At a time when funding for AI infrastructure shows some signs of slowing, the project is a reminder that the scientific enterprise itself depends on sustained investment. Such tools can help ensure that the benefits and resources of AI extend further into the science that makes these advances possible.
The study was led by Dashun Wang of Northwestern University and Nan Cao of Tongji University, with colleagues including Yifeng Wang, Yifen Qian, and Benjamin F. Jones. Their work builds on an emerging field called the “science of science,” which uses statistical and computational methods to study how research itself develops. Read the paper on Arxio at this link.
Related
