Intelligent Descriptive Answer E Assessment System Ideas

Abstract

Evaluation is a vital task in the teaching-learning process. Automatic Short Answer newlineGrading (ASAG) is an ever-increasing realm in natural language understanding. It newlineaims to ease the challenges of teaching, particularly in classrooms with numerous newlinestudents, where assessing brief descriptive responses poses a significant challenge. newlineThe current research work proposes a model answer-based framework for ASAG, newlinenamely IntelliGrader, which addresses three key evaluation facets: i) automatic newlinescoring of short, descriptive answers written in English, ii) identifying the newlineinconsistency in assessment, and iii) providing explicable feedback concerning the newlineinconsistently evaluated answers to the evaluator. newlineHere, the automated scoring is achieved using a model answer-based (reference newlineanswer) approach, which involves performing a collaborative analysis of eight crucial newlinefeatures. Initially, the ASAG task was treated as a regression task, and the models newlinewere built using various state-of-the-art regressors by considering eight independent newlinefeatures and the actual mark given by the evaluator as a dependent feature. The newlineproposed system has undergone thorough validation and testing across four newlinebenchmark datasets: Automatic Student Assessment Prize-Short Answer Scoring newline(ASAP-SAS), SciEntsBank, STITA, Mohler (Texas), and a primary dataset, IDEAS. newlineThe experimental results show the regression results with the finest RMSE of 0.09 newline