Automatic Correction of Free Format MCQ Tests

Automatic Correction of Free Format MCQ Tests

Muaaz Habeek (University of Abdelhamid Mehri - Constantine 2, Constantine, Algeria), Charaf Eddine Dridi (University of Abdelhamid Mehri - Constantine 2, Constantine, Algeria) and Mohamed Badeche (University of Abdelhamid Mehri -Constantine 2, Constantine, Algeria)
Copyright: © 2020 |Pages: 15
DOI: 10.4018/IJSI.2020010103
OnDemand PDF Download:
No Current Special Offers


Although the technology for the automatic grading of multiple-choice exams exists, it is neither efficient nor as automatic as it claims to be. All proposed methods have a predefined answer sheet format that looks like a crosswords table or a chessboard. Because of this format, all questions must have the same number of choices. Such an answer sheet is not clear, and candidates taking the exam can and will accidentally mark the wrong cell in the table. Most of them assume that there is only one possible answer for every question. This article proposes an algorithm that does not require any special format, works with all scanning resolutions and is actually fast.
Article Preview

1. Introduction

Nowadays, nearly everything is computerized, especially dumb, dull or dangerous things, like grading Multiple Choice Questions (MCQ) exam. Automatic MCQ grading is a relatively young research topic. The first system had been developed using Optical Mark Recognition (OMR) forms coupled with OMR software and dedicated scanners (Soumitra et al., 2016), these systems are oriented to big organizations and universities, but small institutes and individual teachers cannot afford these costly systems (Rakesh et al., 2013; Shubham et al., 2015; Remco, 2012).

Automatic MCQ grading systems are based on the extraction of response marks from scanned exam answer sheets. Many methods impose a special sheet format and do not support all types of MCQ such as: conventional MCQ, alternative MCQ and complex MCQ (Fisteus et al., 2013; Abuzar et al., 2016; Bouyy & Leticia, 2016).

Some systems have test generators (Francisco et al., 2016; Nithin et al., 2018) i.e. they provide the ability to generate the forms. This kind of system is often the worst because of its limitations and its strict specifications and conditions (ex: the same special form format for all exams and the candidate identification (id) is not interpreted automatically (Mahmoud & Khaled, 2018) or is written in a complex grid by checking boxes).

Many articles were published in this domain, yet the existing software fails to deliver an easy and practical solution because of their poor image processing techniques (Gokhan, 2017).

In this paper, we present a method that aims to fix all these issues and imposes as few restrictions as possible on the users (candidate / examiner). This method does not require a special sheet format, although it may require certain additions to the answer sheet, such as a rectangular box to contain the candidate id. The candidate id would be written in a more natural way by writing seven-segment digits. This method allows different number of options for questions, for example, the first question may have two options while the second may have twenty. These options do not have to have squares, they can have any type of enumeration (Karandiakr, 2010; Ammar, 2009).

Complete Article List

Search this Journal:
Open Access Articles
Volume 10: 4 Issues (2022): 2 Released, 2 Forthcoming
Volume 9: 4 Issues (2021)
Volume 8: 4 Issues (2020)
Volume 7: 4 Issues (2019)
Volume 6: 4 Issues (2018)
Volume 5: 4 Issues (2017)
Volume 4: 4 Issues (2016)
Volume 3: 4 Issues (2015)
Volume 2: 4 Issues (2014)
Volume 1: 4 Issues (2013)
View Complete Journal Contents Listing