Use of generative AI in requirements engineering

Context

This thesis, which will be written in cooperation with TWT GmbH(https://twt-innovation.de), deals with an LLM-based tool for the automatic formalization of requirements in natural language in standardized templates, so-called boilerplates. A central challenge in requirements engineering is to ensure the consistency and quality of requirements in natural language, which are often written by different stakeholders in different styles. The formalized requirements are used to create a knowledge graph that is readable by both humans and machines and that can be fed back into the LLM to improve the quality of future formalizations.

Objectives

• Building and using a knowledge graph with an LLM
• Training embedding models to select similar requirements
• Fine-tuning of an LLM
• Implementation of an exemplary use case and evaluation of the developed AI method

Requirements

• You have excellent programming skills in Python.
• You have experience in the field of AI, data science or LLMs.
• You have an above-average degree of initiative and a careful, conscientious way of working.
• You have very good written and spoken German and English skills.