Impact of automated short-answer marking on students' learning: IndusMarker, a case study

Abstract

IndusMarker is an automated short-answer marking system based on structure-editing and structure-matching rather than extensive use of linguistic features analysis. Since IndusMarker cannot guarantee 100% human-system agreement rate, the use of IndusMarker has therefore been limited to conducting practice tests. It was expected that such a use of IndusMarker will lead to improvements in student learning and instructor-student interactions. The main aim of this paper is to verify these claims. The results indicate that such a use of IndusMarker leads to improvements in both student learning and instructor-student interactions. In addition, IndusMarker is also shown to give reasonably high human-system agreement rates even after the removal of all linguistic analysis features from the software.