A very large base of students are required to write essays as a means of testing in a lot of countries across the world. From College Board to CBSE, from SATs to board exams, essays have proved to be proficiency tests for grammar, creativity and analysis. With the advent of the SAT exams, many other examinations have taken inspiration from the same on including an essays as a compulsory criteria for admission processes. It is also a known fact that a humongous number of students take these examinations. One of the biggest problems is the evaluation of such essays on such a wide base of students. This not only requires intensive human labour, but is also excessively time consuming.
The solution, we believe, lies in the creation of a powerful system that automates Essay Grading. Essays mainly stand out on the choice of words and how relevant they are to the topic at hand. So, we plan on introducing a technique that provides a higher weightage (or score) to these words and also a score as to how they sound grammatically and how they compare themselves with the standard essays out there. Our next step would be to create a neural net that is fed with a lot of essays which would eventually give it a score accordingly.
Our main objective is to reduce both, time and manual labor which goes into the whole evaluation process. Instead of a human, this model corrects and grades these essays accordingly. The dataset for the same has been taken from the Automated Student Assessment Prize dataset (ASAP) which was part of a Kaggle competition hosted by the Hewlett Packard Foundation.
To make the UI we have used an open source app framework called Streamlit. It is fast, easy and extremely user friendly!
//to launch the web App you need to enter "streamlit run <name of python file with .py extension>" in the command line. 1