We are accessing ChatGPT and Github Copilot API to compare their performance when recognizing design patterns Singleton and Observer. Moreover, to see the effect of prompt styles Context Manager and Persona on the LLMs. This is done as part of our Bachelor Thesis in Software Engineering and Management at Gothenburg University.
This repository consists of helper scripts that access Chatgpt and Github Copilot API for automation of data collection for our Bachelor Thesis. Our thesis aims to evaluate the extent to which these LLMs can recognize design patterns (Singleton and Observer) and the effect different prompt styles (Persona and Context Manager) may have on the LLMs. The scripts run defined prompt examples of Persona and Context Manager along with the code examples of Singleton and Observer which were taken from open-source projects found on PMART (https://www.ptidej.net/tools/designpatterns/). The data collected with this script is saved in CSV files for us to analyze for our thesis.
- How/where to download your program
- Any modifications needed to be made to files/folders
- Dimitrios Pokkias
- Shahd Metwally
This project is licensed under the [NAME HERE] License - see the LICENSE.md file for details
Inspiration, code snippets, etc.