Identifying personal health information using support vector machines

We explore the use of Support Vector Machines to recognize personal health information in medical discharge summaries. In addition to the basic token level features, we use entities recognized by an information extraction system designed for newswire text, plus a set of rules that incorporate entityspecific knowledge. The results on the unseen test dataset show that the SVM model can be easily adapted to a new domain with minimal work and achieve good performance (0.9869, the weighted F measure). The proposed new features also contribute to improving the accuracy of entity identification.