----------------------------------------------------------------------- BIOINFORMATICS COLLOQUIUM School of Computational Sciences George Mason University ----------------------------------------------------------------------- Information Extraction for Biomedical Science Victor Pollara Mitretek Systems Tuesday, October 26, 2004 4:30 pm Verizon Auditorium, Prince William Campus With the number of completed clinical trials already exceeding 300,000, how does one find information and how does one evaluate the results, the validity, and the usefulness of a clinical study given this information overload? Information retrieval (i.e. document indexing) is a crude tool used to search for documents containing complex biological and medical information. But it presents users with whole, uninterpreted documents which they must read to find the sought-after facts. Information Extraction (IE) is revolutionary in the sense that its purpose is to locate, extract, and organize the facts contained in documents. IE and the ontological data modeling needed to support it are complex problems, and the technologies for addressing them are not yet mature, but increasingly sophisticated commercial and academic software has brought the prospect of practical, ontologically driven IE close to hand. This talk describes an ongoing research project at Mitretek Systems, in which the goal is to integrate cutting edge natural language processing, information extraction, and ontological modeling of biomedical knowledge in order to extract, organize, and analyze the biomedical information in clinical trials reports, and to provide the world of biomedical research with an unprecedented resource for analyzing the extracted information. ---------------------------------------------------------------------- Refreshments are served at 4:00 pm. Find the schedule and directions at http://www.binf.gmu.edu/colloq.html