Test to Code Traceability

This paper presents a novel approach to establishing test-to-code traceability, using requirements as a mediator between functional tests and source code. Our approach consists of two steps: first, we link functional tests to requirements, and then we use these links to establish connections between requirements and code. We implement our approach using a combination of traditional and advanced machine learning models, including cosine similarity and GPT-LLAMA 2. By leveraging these models, we are able to automate the traceability process and ensure that tests are accurately linked to the relevant code components. Our implementation demonstrates the feasibility and effectiveness of our approach in establishing and maintaining traceability between functional tests, requirements, and source code.