Envoc and LinkIt! Achieve 99.2% Accuracy on Bubble-Sheet Grading with Software
Automated grading with near-perfect accuracy was delivered with LinkIt!’s “Intelligent Bubble Sheet Grader” software developed by Envoc, a software and mobile app development company.
The intelligent bubble-sheet grader software allowed for the grading of bubble sheet tests that were previously rejected, by integrating Atalasoft’s DotImage software and achieving a 99.2% success rate. Tedious, manual grading was the only option available to grade the thousands of machine-rejected bubble test forms. With LinkIt!’s solution, grading speeds increased to a rate of 80 per minute, with staggering accuracy.
Getting accurate and timely test results for a large student population is critical in assessing success and course impact in the classroom. “These days, educators have less time and fewer resources to serve the needs of students,” according to Joshua Powe, LinkIt!’s co-founder and president. “When coupled with an increased focus on accountability, the importance of getting accurate data in a timely fashion has never been greater. Creating solutions that help drive student achievement through the effective use of technology has always been a focus at LinkIt!, but our work with Envoc has raised the bar of what is technically possible. More efficient software for processing bubble sheets means more instructional time for students, and ultimately, accelerated learning gains.”
Machine-graded tests fail the student when they don’t account for unrecognized marks, erased answers, scribbling, and what appear to be multiple attempts at the same question. By leveraging innovative software tools and technologies, discerning the intent of the student on standardized tests by a computer can be achieved with the same rate of accuracy as a human grader. “Teaching a computer to discern the intent of K-12 students who scribble and mark up their bubble sheet exams was a challenge,” said Matt Vidacovich, Developer at Envoc, “Their artwork, however, is no match for the image handling handiwork of Atalasoft’s DotImage, peppered with an applied Neural Network.”
Teaching a computer to discern the intent of K-12 students who scribble and mark up their bubble sheet exams was a challenge
MATT VIDACOVICH, DEVELOPER AT ENVOC