Masters Project: Utilising LLM agents for requirement analysis and specification from preliminary meeting notes

Project Overview

Aim

The primary aim of this dissertation is to investigate the effectiveness of LLM agents in aiding requirement analysis and specification from preliminary meeting notes in software engineering projects.

Objectives

1. Evaluate Efficiency and Accuracy: To assess the efficiency and accuracy of multi-agent system powered by different LLMs in requirement analysis and generated requirement specifications. This also involves if the business requirement specifications generated is GDPR-compliant and a comparative analysis of these generated documents across the different LLM utilized.
2. Measure Quality and Time: To evaluate the quality of outputs from agentic workflows and the time taken for requirement analysis and specification processes. This entails assessing the quality of requirement specifications produced by LLM agents using established and predefined criteria via human evaluation and LLM-as-a-Judge. This will provide insights into LLMs' operational benefits, time-saving potential and user satisfaction in requirement engineering.
3. Develop and Experiment: To develop an application that integrates the various components of the proposed agentic system and using version control system to track changes in the Application Programming Interface (APIs), source code, prompts and outputs across various experiments. This involves creating a user-friendly interface and ensuring seamless interaction between the multi-agents system and users.
4. Complete Dissertation and Product Development: To complete the dissertation writing and product development within the stipulated period. This ensures that all research activities and reporting are completed within the academic timeline. The dissertation will be written in parallel with product development, ensuring both the written document and the functional prototype are finalized by the end of the dissertation period. Periodic reviews and revisions will maintain the quality and relevance of the research output.