Towards Automating Database Designing

Database designing is an important process for enabling good quality data. Without designing the database correctly, the database might contain the same data several times, or it might contain data that is not usable for decision making. The evolution of software development, programming languages, increasing amount of data, different data models, different data sources and many more have increased the importance of designing databases to provide accurate data for decision making. Designing databases manually is time consuming. If the process can be automated, it would allow faster creation of good quality databases. The goal of this study is to investigate whether large language models could be used for designing a Data Vault 2.0 raw database to automate the designing process. In this study we introduce database designing as a process, and describe the main principles of Data Vault 2.0. We create an example data source, an example Data Vault 2.0 raw database based on the source database for reference, and then test the ChatGPTs capabilities for creating a Data Vault 2.0 raw database based on instructions given in a prompt. Finally, we analyze the results and discuss future works.