NSF hopes big data will finger grantees not reporting foreign support

The National Science Foundation (NSF) will soon begin crunching several large databases to see whether there are scientists who failed to disclose ties to foreign institutions in their grant applications. It is arguably the boldest of several steps federal research agencies are taking to comply with https://www.science.org/content/article/what-big-new-u-s-law-reshapes-science-agencies-could-mean-researchers”>a new law that aims to boost U.S. technological innovation–and prevent China and other foreign governments from pilfering federally funded research.

The CHIPS and Science Act, signed by President Joe Biden on 9 August, appropriates $52 billion over 5 years to stimulate research, training, and manufacturing in microelectronics—and promises tens of billions more for fundamental research in many fields. Along with those investments is a mandate to strengthen research security (see #sidebar”>sidebar, below).

In recent years, lawmakers and others have faulted federal research agencies for failing to be more vigilant against potential security problems. The list includes situations in which grantees have accepted foreign funding that comes with restrictions on publication or that creates a “conflict of commitment” for a scientist employed by a U.S.-funded entity. China’s aggressive recruiting of U.S. scientists, many of Chinese ancestry, has been a particular concern.

In June 2018, the National Institutes of Health began to https://www.science.org/content/article/has-it-peaked-i-don-t-know-nih-official-details-foreign-influence-probe”>more aggressively enforce existing rules requiring grantees to disclose any foreign ties, resulting in sanctions against some scientists and the return of some grant funds. Five months later, the U.S. Department of Justice launched the China Initiative, a law enforcement campaign to thwart economic espionage by the Chinese government. It has resulted in the prosecution of some two dozen academic researchers with links to Chinese institutions. In recent years research units within the departments of energy and defense have created what they call a “risk matrix” to help identify potential security threats to research they are funding.

The CHIPS act enshrines some of those practices into law as well as requiring agencies to assess the types of research most vulnerable to theft, provide more training to scientists on how to reduce security risks, and gather more information from grantee institutions. It also bars scientists employed by the U.S. government from joining a talent recruitment program run by any other country and prohibits federal grantees from participating in talent programs funded by government entities in China, Russia, Iran, and North Korea.

At NSF, officials have decided to turn to big data to help safeguard the agency’s $7 billion research portfolio. The agency already reviews the biosketches that accompany each grant proposal and provide basic information about each applicant and key members of their team, including institutional affiliations, collaborations, areas of research, and geographic location. Going forward, NSF will compare what applicants have disclosed with information contained in two commercial databases of scientific publications—the Web of Science and Scopus—as well as U.S. patent applications.

NSF says its goal is to spot potential red flags, including omissions or inconsistencies that could violate its policies. Of particular concern would be an NSF grantee who has listed participation in a foreign talent recruitment program in a published paper—but not disclosed that tie to NSF.

“Very often the researcher will acknowledge one of these talent plans in their paper because it’s a requirement in their contract to do so,” says Rebecca Keiser, head of NSF’s office of research security. “So now we’ll be able to find that through data analytics.”

NSF will take a closer look at any discrepancies it finds, Keiser says, and then reach out to the researcher’s institution for more information. (To this point, NSF program staff has referred individual cases to the agency’s independent inspector general, who then decides whether to investigate.) “We will ask an institution to work with us to understand whatever it is we have found,” she says.

University administrators first got wind of NSF’s plans in November 2021 when the agency put out https://www.govinfo.gov/content/pkg/FR-2021-11-09/pdf/2021-24487.pdf”>a public notice of its intention to create a new “system of records.” But NSF has yet to spell out exactly what information it will collect and how it will manage those data.

That has prompted some anxiety among academic researchers. The Council of Governmental Relations (COGR), which tracks the impact of federal regulations on academic research for its 200-plus member institutions, has expressed concern about who would have access to the data files and how NSF would validate their accuracy. “We’re still waiting to learn the rules of the road that apply to this new system of records,” says COGR’s Kristin West.

COGR would like institutions to have the chance to vet any discrepancies that NSF finds before the agency begins to ask questions. But NSF’s Keiser says that won’t be possible, because the information it collects from grant applications is confidential.

NSF plans to make its data-mining algorithms available to institutions so they can do their own analyses and resolve potential disclosure issues before they come to NSF’s attention. “It’s a research tool, and we want everybody to have access to it,” Keiser says.

That tool could also be a boon to scientists, Keiser says, by helping them identify other groups doing similar research and opening the door to possible collaborations. As Keiser imagines it, “We might call the university and say, ‘Hey, the analytics found this really high-impact project you may not know about. Isn’t that awesome?’”

Keiser estimates she needs four people to carry out the analytics, do the “human validation,” and then interact with the academic community, although she notes that her office’s current budget is insufficient to meet that staffing level, which is mandated by the CHIPS act. Even so, Keiser says, a “creative use” of existing resources may allow her to launch the data mining project before the end of the year.