A Proposal of a Method to Extract Active Ingredient Names from Package Inserts

Recently, medical accidents caused by drugs have attracted attention in Japan. To prevent this from affecting medical experts, it is effective to improve the `safety of drug usage' via information systems. Although package insert information in SGML format is provided by PMDA, it is not easy to utilize this information for such systems because they lack a suitable structure to provide direct access to the required information. In this study, we propose certain methods to extract the active ingredient name from such unstructured SGML data.