Scoring Free-Responses Automatically: A Case Study of a Large-Scale Assessment
Claudia Leacock · 2004
C-rater is an automated scoring engine that measures a student’s understanding of content material through the use of natural language processing techniques. We describe the process used for building c-rater models using Alchemist, c-rater’s model-building interface. Results are given for a large-scale assessment that used c-rater to score 19 reading comprehension and five algebra questions. In total, about 170,000 short-answer responses were scored with an average of 85 % accuracy. 1.