Measuring ChatGPT Efficiency
This project has two goals. Our first goal is to evaluate the efficiency of ChatGPT in answering questions from various domains such as History, Social Science, and Computer Security, while also measuring its response time and accuracy .
Our second goal is to design and implement an Interactive Website that will document our research, integrate data processing, display data visualizations, and provide insight on ChatGPT's performance.
Research Approach
To evaluate ChatGPT's efficiency in responding to domain-specific questions, these are the tools, technologies, and libraries used:
- GitHub Repository: used to host web content and store project files.
- GitHub Desktop: used to sync changes to the repository.
- MongoDB Atlas: cloud-based database used to store questions, ChatGPT's responses, and response time.
- MongoDB Atlas Charts: data visualization (pie charts, bar charts, etc.).
- OpenAI API: for sending ChatGPT our questions and retrieving its answers to be used for evaluation.
- Node.js: for server-side environment creation.
- Express.js: a web framework for Node.js.
- Mongoose: a library to define schemas.
Dataset Overview
Our dataset consists of questions from three different domains:
- History
- Social Science
- Computer Security
Each question includes fields such as:
- _id
- question
- options (A, B, C, D)
- correctAnswer
- chatGPTResponse