A study of computer-aided grading of written team work

Abstract

The trend towards increased enrolment in higher education has inevitably led to larger classes and increased the stress on educators including the demand on their time and budgets. One of the most significant effects of this is the increased time and effort spent on grading and verification of discrepancies between different graders. A possible aid to deal with this problem is to provide instructors and teaching assistants with computer­‐aided means of analyzing the written work. The purpose of this study was to explore the use of a computational technique called “Latent Semantic Analysis” (LSA), to aid student authors, their teaching assistants, and instructors in the assessment of written project work in design engineering. Several tests were performed on the course team documents and assignments, to compare LSA’s reliability to human graders for accurate and effective grading. LSA produced accurate results and proved to be more reliable than an average grader for the set of documents tested from the engineering course, ENGINEER 1P03 – Introduction to Professional Engineering.