Current Awardee

Identifying Conflicting Claims in Clinical Literature using Natural Language Processing and Knowledge Graphs

Halil Kilicoglu
Halil Kilicoglu

College: School of Information Sciences
Award year: 2021-2022
NCSA collaborators: Michael Bobak, Colleen Bushell

A vast amount of scientific knowledge is published in biomedical publications on a daily basis. Some of these publications make new knowledge claims, while others confirm or refute earlier ones. Information retrieval tools are routinely used to locate and retrieve publications relevant to user queries; however, it is still largely up to the reader to interpret the specific claims made in a publication and contextualize them to assess their significance, robustness and credibility. With the size and the exponential growth in the biomedical literature, this is a challenging and time-consuming task. While natural language processing (NLP) techniques are increasingly used to extract information from scientific publications, little attention has been paid to specifically interpreting knowledge claims and situating them within the larger scientific evidence base. In this project, the researchers will to redesign an existing biomedical NLP tool, SemRep, to more effectively extract knowledge claims and their contextual characteristics from scientific publications and to generate publication-specific knowledge graphs. The team hypothesizes that modeling the claims in a publication and their contextual characteristics as a knowledge graph could enable the search, verification, and tracking of scientific claims at literature scale and underpin automated literature surveillance tools.