Web application for Generating Entity-Relationship Diagrams with Natural Language Processing

Abstract

Entity-Relationship (ER) diagrams are widely used in software engineering to model relational databases and serve as semantic representations easily understood by humans. They convey the meaning of data, their relationships, and key modeling concepts. Despite their usefulness, designing ER diagrams requires technical expertise. This work proposes a prototype leveraging natural language processing tools, particularly large language models (LLMs), to simplify and democratize their creation. Using simple prompts, the system generates structured outputs in standard formats that can later be used for ER diagram construction. As a result, a web application was developed, offering a user-friendly interface that reduces time and effort, aiming to make diagram creation more accessible.