AI for Legacy Code

Created on

Call for solution type:
SpeedUP
participation dates:
from 07-10-2024 to 11-11-2024
StatuS:
Open for applications
Deadline for QUESTIONS:
06-11-2024
Optional information meeting:
15-10-2024
CONTACT:
cciop@digital.etat.lu
Full Description EN (Pdf, 236 Kb) Descriptif complet FR (Pdf, 215 Kb)

The French version of the description was updated on October 15, 2024. 

Name

AI for Legacy Code

Ordering party

Ministry for Digitalisation

Types

Production of technical and functional documentation for software source code

Objectives

The MM-AET tool of the Ministry of the Economy is an application for managing establishment permits and financial aids. The tool, coded in Java, has evolved significantly over the years, and the documentation is not always complete or extensive, making the work increasingly complex when adding new features or debugging, due to the time required to understand the connections and impacts between various parts of the code.

The purpose of this challenge is to use artificial intelligence technologies to document the source code of the MM-AET software comprehensively and intelligently. The idea is to use an AI system to read and understand the source code, conduct in-depth syntactic and semantic analysis to extract the structures, architectures, and links, to generate functional and technical documentation.

Challenge details

 

The project expectation is the creation of detailed technical documentation using AI analysis tools, including:

  • Description of functions, classes, and modules
  • Explanation of control flow and algorithms
  • Identification and explanation of dependencies
  • Original source code enriched with comments
  • Generation of UML diagrams (classes, sequences, components) and architecture diagrams
  • Production of a contextual technical glossary
  • Analysis report on code structure and complexity
  • Code quality analysis and improvement suggestions
  • Traceability of links between different parts of the code

 

Technical specifications of the software:

  • Code in Java – approximately 700,000 lines

                    Java 8 (limited version with Websphere 9)

                    Main frameworks and libraries used: see Appendix

                    Front end based on Struts, JSP Page, jQuery, Vue.js technologies and frameworks

  • Versioning in GitLab 17.1 (CTIE version)
  • Database in DB2, 134 tables

 

For complete information regarding the requirements, the deliverables and the type of offer to be submitted, please refer to the document linked on this page.