Abstract
It is very useful to constructing system that can predict future situation of illness from medical records written by patients in natural language because it will help to ensure informed consent. As the first step, we propose a method to extract description of the effect caused by taking the medicine as a triplet of expressions - medicine name, object of change, and its effect from illness survival blogs. There is no database of evaluation expressions for medical treatment. Moreover usual extraction patterns are not suitable for processing these blogs. Therefore, we decided to extract a expressions triplet using specific clue words and by parsing results. In the experiments, we have confirmed that medication usage information can be extracted with high accuracy compared to other existing methods.
