Ontology Development Information Extraction (ODIE)

Rebecca Crowley, University of Pittsburgh

This collaborative R01 is exploring the use of Natural Language Processing techniques to create an open-source platform providing researchers a workbench for extracting data to enrich ontologies, and using ontologies to do said extraction.

This toolkit performs information extraction tasks – such as named entity recognition, co-reference resolution, extraction of attributes and value, and discourse reasoning – when using ontologies from BioPortal to extract information from clinical documents. It then completes the cycle by performing ontology enrichment tasks such as discovering new concepts and synonyms, and suggesting taxonomic placement and relationships using information gleaned from clinical documents.

ODIE is a research project at the Department of Biomedical Informatics at the University of Pittsburgh, currently being developed as a collaborative R01. It employs Natural Language Processing techniques to identify and retrieve relevant free text information from clinical document repositories using ontological terminology, with the goal of improving and enriching ontologies, thus supporting the cycle of information transfer.

Learn more About the ODIE Toolkit, or Watch a Demo.