The problem CheckMate solves

In case of online submissions in the form of pdf files, etc. It becomes tedious as well as a difficult task for professors to go through all submissions and check for plagiarism and cheating.
With this ML-powered tool, this task can be done with the ease of a click.
Also helps the user to perform an auto evaluation of the submission documents with a reference document.

Challenges we ran into

1. Cleaning textual data: We developed a method to tackle the repetition of textual data in the files.
2. displaying of results: We got the similarity scores in a sparse matrix but we altered it to get the final result in a string format, with the respective pdf names.
3. Dealing with multiple files simultaneously: Generally only 2 files are compared, but we devised an algorithm to compare each document with another.
4. Integrating the web apps: We created a flask API to connect the frontend with the backend model to display the expected results.
5. Accuracy of the auto evaluator: The rate of decrease in the similarity score was too steep. So we generalized the final result to make it more realistic.