Using Active Learning Methods to Strategically Select Essays for Automated Scoring

Research on automated essay scoring has become increasing important because it serves as a method for evaluating students' written-responses at scale. Scalable methods written responses are needed students migrate to online learning environments resulting in the need evaluate large numbers of written-response assessments. The purpose this study is describe and three active than can be used minimize number essays that must scored by human raters while still providing data train modern system. uncertainty-based, topological-based, hybrid method. These were select included part Automated Student Assessment Prize competition then classified using model was training with bidirectional encoder representations from transformer language model. All produced strong results, topological-based producing most efficient classification. Growth rate accuracy also evaluated. different levels efficiency under sample size allocations but, overall, all highly classifications similar one another.