NSF awards professor grant to study potential role of non-researchers in advancing scientific research
Since the aforementioned project would require significant advances in AI, Doroudi proposed a complementary, two-phase approach that relies on human intelligence.
Phase 1 is an exploratory phase to understand the differences in ability between groups of non-researchers and researchers to complete literature review tasks, determine the ability of non-researchers to find information among different scientific topics, assess training methods, and assess the effects of creating cohorts of individuals that show promise in a specific area of science.
In Phase 2, a field study will determine how well the findings in the first phase work when employed in real-world use case literature searches.
“The project will assess the viability of creating a paid workforce of trained workers that can help researchers conduct scientific literature searches, but it will also result in a platform to involve interested high school students and undergraduates in this process as citizen scientists,” Doroudi explained.
EAGER grants fund projects that NSF considers "high risk – high payoff."
Doroudi believes that his EAGER project will prove a novel means for the public to engage with scientific research, and that his research findings should contribute to advances in two additional NSF Idea Machine topics, “Creating Sustainable Education Pathways” and “Reinventing Scientific Talent.”
As part of the project, Doroudi plans to specifically target and train high school and undergraduate students who are historically underrepresented in STEM.
Doroudi’s research builds on his early work as an intern at Microsoft Research, where he trained crowdworkers to perform complex web search tasks. In subsequent research projects, he recognized the importance of finding related work coming from other disciplines or work that was conducted decades ago.
“When I tried different methods of training crowdworkers, I noticed that at times they could complete these tasks more successfully than researchers, and that their performance could improve with small amounts of training.”