Large Language Model-Based Database and Form Automation
Bildiriler > Tam Metin Bildiri
Yazarlar: TUGAL, SAMI, Murat Osman ÜNALIR
Zenodo (CERN European Organization for Nuclear Research), , , Uluslararası, Tam Metin Bildiri
Özet
This study proposes a feedback-driven, interactive Large Language Model (LLM) based system that converts user requirements expressed in natural language into both a relational database schema and executable form interfaces through a single intermediate representation. Using commercially available large language models, the system carries out requirement elicitation, schema generation and form personalization in an integrated pipeline based solely on natural language descriptions, without requiring the drag-and-drop interaction demanded by traditional Low-Code/No-Code (LCNC) platforms. A multi-agent architecture provides an iterative dialogue mechanism for resolving requirement ambiguities, while user feedback is processed at the intermediate representation level so that database and interface outputs are updated consistently. The developed prototype was evaluated on a library management scenario, and the system was observed to generate an executable database schema together with form components directly from natural language input.