Development for DB schema reconstruction support tool using natual language processing

Abstract

When computer systems are used for a long time, a database to be used is changed due to various reasons and the database schema are also modified accordingly. However, the schema modification works depend on the experience and skill of engineers and cause many problems. In this paper, we describe a tool that supports the modification of the database schema. This tool is based on the assumption that table names, column names, and column names belonging to the same table are closely related when an engineer develops an appropriate database design according to the data to be managed. We proposed a method to restructure a database schema based on distances among table names and column names, and implemented it as a tool. We applied the tool for the actual database schema to verify the above assumption and showed effectiveness of the tool.