Bei Yu, Professor at the School of Information Studies, Syracuse University
Friday, December 6, 2024
1:30pm
Luddy Hall, 4063
https://iu.zoom.us/j/81502757332
An NLP Analysis of Causal language Use in Scientific Communication
Abstract:
The prevalence of drawing causal conclusions from observational studies has raised concerns about exaggerated claims in both scientific papers and news. This talk introduces our work in developing NLP models to automatically identify correlational, causal, and conditional causal claims in research papers and press releases. These models were then applied to examine patterns and trends of causal language use by scientists and journalists. The result shows that 22% of press releases made exaggerated causal claims from correlational findings. University press releases exaggerated more often than those from journal publishers. Encouragingly, the exaggeration rate has slightly decreased over the past 10 years, despite the increase of the total number of press releases. Among scientists, The use of causal language seems to vary: women authors, authors with more research experience, authors from larger teams, and authors from countries with a culture of uncertainty avoidance, are less likely to make causal claims when reporting results from observational studies. We will discuss the implications of these findings and explore the potential of using LLMs to monitor causal language use in scientific communication.
Bio:
Bei Yu is a Professor at the School of Information Studies, Syracuse University. Her research focuses on using machine learning and natural language processing techniques to assist scholarly communication and science communication, especially in monitoring and improving the quality of scientific information.
Luddy calendar event: https://events.iu.edu/siceiub/event/1692736-an-nlp-analysis-of-causal-language-use-in
Leave a Reply