December 5 @ 4:00 – 5:00 pm
Speaker: Stephanie W. Haas, PhD
Presented from UNC-CH
Broadcast Link: Seminar
Information classification tasks require sorting records into one or more categories, usually based on the comparison of document features to category definitions. Classification systems combine human expertise and machine learning techniques to meet users’ performance requirements, while dealing with the constraints imposed by the documents and definitions. Syndromic surveillance identifies potential outbreaks of health problems such as Gastro-Intestinal illnesses by classifying patient records that are likely to represent a patient with that illness. Legal discovery identifies letters, email, and other documents that are likely to be relevant to a specific legal matter. Human expertise is an expensive and valuable resource. In this presentation, I compare these classification tasks to explore the interaction between task characteristics and the use of human expertise.
Dr. Haas is a professor in the School of Information and Library Science at UNC Chapel Hill, and Coordinator of the MS Information Science Program. Her research interests include information representation in medical records and the coordination of information representation and work processes in health care. Several research projects in collaboration with colleagues at the North Carolina Disease Event Tracking and Epidemiologic Collection Tool (NC DETECT) have focused on concept extraction from the chief complaint and triage note fields in Emergency Department patient records, and their use in syndromic surveillance. An award-winning teacher, Dr. Haas regularly teaches courses in Applications of Natural Language Processing, Database Design, and Information Systems Analysis.