| KEYWORD |
LLM to generate code from formal requirements
Reference persons RICCARDO COPPOLA
Research Groups DAUIN - GR-16 - SOFTWARE ENGINEERING GROUP - SOFTENG
Description The thesis involves the selection, customization, integration, and training of a generative AI - in the form of a Large Language or Foundation Model - for code generation from use formal requirements. The task includes a first phase in which a thorough foundation and understanding of various use case languages, and formal languages for the description of requirements. The task will also include the definition of modules and tools for the translation of natural language and unstructured requirements towards structured and formal notations (e.g., Use Case Diagrams and Narratives).
Activities will include data preprocessing, where diverse datasets will be collected, cleaned, and structured. Generative AIs will then be trained and fine-tuned using these datasets to capture the nuances of different use case scenarios. Continuous evaluations and refinements of the AIs will be necessary to ensure correctness.
Deadline 06/12/2024
PROPONI LA TUA CANDIDATURA