Automated essay grading with recommendation

Abstract

In our thesis we have worked to analyze text essays then predict the score accordingly and also recommend similar essays as well as other noticeable required changes to the readers who want to improve their essay writing skills. In our research we have used a dataset of 13000 essays scored by two human graders provided by the Hewlett foundation available in Kaggle. We have used different natural language processing techniques and enormous natural language tools and tried to see different patterns present in the essays to score them. We have extracted noticeable features from these essays created dataset with necessary formation then again used supervised machine learning models to build an artificial system that could score further user given essays and also make suggestion. We have implemented a machine learning agent which is trained by linear regression algorithm on the extracted features to predict the score and then calculates cosine distance to determine similar helpful essays and recommends those essays to the users. Also we have developed our system to suggest the writer necessary correction of their mistakes and writing patterns.