As the body of legal texts grows the ability to extract context-relevant information automatically becomes more important. In this work we extract pre-specified types of information related to emergency preparedness and response of public health system from Pennsylvania public health statutes. For this task we use a dataset created by policy analysts at the University of Pittsburgh's Graduate School of Public Health. They extracted the information from the statutory texts manually and represented it in the form of numeric codes in a coding table.
Preliminary results obtained in the past suggest that supervised learning is a promising approach to extract the information automatically. In current work, we are generating additional features to represent the texts and using a more refined approach to the classification task based on additional task analysis. We have decomposed the task into smaller tasks that may be tackled individually in an attempt to improve the performance of our classification framework.