Application Of Machine Learning For Automatic Scoring Of Afaan Oromo Subjective Exams

Abstract

Answer Scoring Is An Essential Part Of The Student Evaluation Process In The Education System. In An Exam, Students Need To Answer Subjective And Objective Questions. In Educational Institutes, Instructors Need To Evaluate The Answer Script Manually To Evaluate The Students. Number Of Students Enrolled In Different School Throughout The Country Increases In Higher Rates Every Year. An Increase In Student Number Leads To An Increased Demand For More Teachers And Student Assistants, And For Every Exams Given Through The School, More Graders. To Resolve This Need, This Thesis Explored The Possibility Of An Automated Subjective Answer Scoring System. The Proposed Method Learn Scoring Patterns From Human Graders By Extracting Features From The Student Answer And Using Them To Train A Machine Learning Algorithm To Score Another Student Answer Easily. We Intend To Look At How Automatic Answer Scoring Can Be Done By Using Machine Learning Methods With Similarity Between Model Answer And Student Answer For A Given Dataset.The Study Is About Implementing A Machine Learning Method To Automate Scoring Of Subjective Answer By Comparing Student Answer With The Model Answer With Similarity Metrics. We Present Experimental Results On A Dataset Provided From Afaan Oromo Subject In The Meta Walkite Preparatory School. We First Apply Feature Extraction To Both Student Answer And Model Answer And Measuring Similarity Between Them. Similarity Calculation Is Based On The Number Of Common Words. Then Evaluate The Relation Between The Similarities And Marks Awarded By Scorers Using Linear Regression.To Evaluate The System Regression Metrics Are Used. The Experimental Results Show That For Answer Scoring Using Cosine Similarity Between Model Answer And Student Answer We Got Result Of 11% MSE, 33 % RMSE And 27% MAE.