PaperCheker

Add your solution key and answer sheet of students
The report gets generated in few seconds with marks obtained and percentage
click on Process PDFs

Inspiration and Motivation:

The inspiration behind PaperChecker stems from the challenges faced by educators in the field of education, particularly in the process of grading student assignments. The manual grading process is time-consuming, labor-intensive, and prone to human error. This project aims to address these challenges by leveraging the power of artificial intelligence (AI) to automate the grading process, thereby saving time and ensuring a more accurate assessment of student work.

Technology Used:

PaperChecker utilizes two cutting-edge AI models: a sentence transformer model from Hugging Face and OpenAI's Turbo-GPT-3.5. The sentence transformer model is employed to convert text into a numerical representation that captures the semantic meaning of the text. Turbo-GPT-3.5, on the other hand, is used to generate a similarity score between the student's answer and the solution provided. By averaging these similarity scores, PaperChecker calculates a final score for each student's assignment. A key innovation in PaperChecker is the use of the 'few-shot learning' technique with Turbo-GPT-3.5, which allows the model to adapt to new grading tasks with minimal examples, significantly enhancing its versatility and accuracy.

Building the Project:

The development of PaperChecker involved several key steps:

Data Preparation: The initial challenge was to prepare the data in a format that could be processed by the AI models. This involved converting PDF documents containing student answers and solutions into a text format. Model Integration: Integrating the Hugging Face sentence transformer model and OpenAI's Turbo-GPT-3.5 into the project required careful consideration of the models' capabilities and limitations. The sentence transformer model was used to convert the text into a numerical representation, while Turbo-GPT-3.5 was used to generate similarity scores. The application of the few-shot learning technique with Turbo-GPT-3.5 was a critical step in enhancing the model's performance. Scoring Algorithm: The core of PaperChecker is the scoring algorithm, which calculates the final score for each student's assignment. This involved averaging the similarity scores generated by the AI models to produce a comprehensive assessment.

Challenges Encountered:

One of the main challenges was ensuring the accuracy and reliability of the AI models in generating similarity scores. The models needed to be fine-tuned to accurately reflect the nuances of student answers and solutions. Additionally, integrating these models into a cohesive grading system required careful planning and execution. The application of the few-shot learning technique presented its own set of challenges, as it required careful tuning to achieve optimal performance.

Learnings and Takeaways:

Through the development of PaperChecker, we learned the importance of leveraging AI to automate and enhance educational processes. The project also highlighted the challenges and opportunities in integrating advanced AI models into practical applications. The experience gained from this project will be invaluable in future endeavors, particularly in the field of education technology.

Future Enhancements:

Looking forward, we plan to enhance PaperChecker by incorporating additional AI models to improve the accuracy of grading and expanding its capabilities to support a wider range of educational materials. We also aim to develop a user-friendly interface that allows educators to easily input data and receive grading results.

PaperChecker represents a significant step forward in automating the grading process, offering a solution that is not only efficient but also promising in terms of accuracy and reliability. The integration of the few-shot learning technique with Turbo-GPT-3.5 is a testament to the potential of AI in education, showcasing how technology can transform the way we assess and learn.