Next-Gen QA

The full title of the undergraduate research for this project is "Next-Gen Website QA: Automated Functional Test Case Creation Using Large Language Models, Prompt Engineering, and Scraping Technologies". This was my group's thesis during our fourth year in college and it is related to artificial intelligence and quality assurance. The project files are not public, but I can talk about the features in terms of concepts.

This was meant to assist quality assurance devs or experts in creating test cases, by providing a good scaffold of automatically built test cases ready for manual testing. We were able to create a prototype that worked well, but we chose to make the repository private for now.

For the model, we used a fine-tuned Llama3.1 using Unsloth for training and Ollama for the local Llama3.1 model and patched it with the use of a Modelfile.

Features

  • - Scrapes basic interface data (such as headers, buttons, paragraphs) using Selenium

  • - Creates prompts using the scraped data

  • - Generates website functional test cases using the prompts

  • - Outputs a .csv or pdf file containing the test case data.

Credits

I'll keep it broad, for some it is for privacy purposes:

  • - My groupmates

  • - The testers of the project

  • - Numerous references and learning resources on Django, Ollama, Unsloth, and other libraries/tech

  • - Related Research Studies

  • - Some suggestions from Copilot (and doing a lot of cross-checking)