Foundations of statistical natural language processing
Foundations of statistical natural language processing
Beyond the Short Answer Question with Research Methods Tutor
ITS '02 Proceedings of the 6th International Conference on Intelligent Tutoring Systems
A spelling correction program based on a noisy channel model
COLING '90 Proceedings of the 13th conference on Computational linguistics - Volume 2
Assessing creative problem-solving with automated text grading
Computers & Education
Text-to-text semantic similarity for automatic short answer grading
EACL '09 Proceedings of the 12th Conference of the European Chapter of the Association for Computational Linguistics
Automatic short answer marking
EdAppsNLP 05 Proceedings of the second workshop on Building Educational Applications Using NLP
The WEKA data mining software: an update
ACM SIGKDD Explorations Newsletter
Automating Model Building in c-rater
TextInfer '09 Proceedings of the 2009 Workshop on Applied Textual Inference
Language models as representations for weakly-supervised NLP tasks
CoNLL '11 Proceedings of the Fifteenth Conference on Computational Natural Language Learning
Natural Language Processing (Almost) from Scratch
The Journal of Machine Learning Research
| Hi-index | 0.00 |
The SAVE Science project is an attempt to address the shortcomings of current assessments of science. The project has developed two virtual worlds that each have a mystery or natural phenomenon requiring scientific explanation; by recording students' behavior as they investigate the mystery, these worlds can be used to assess their understanding of the scientific method. Currently, however, the scoring of the assessment depends either on manual grading of students' written responses, or, on multiple choice questions. This paper presents an automated grader that can combine with SAVE Science's virtual worlds to provide a cheap mechanism for assessments of the ability to apply scientific methodology. In experiments on over 300 middle school students, our best automated grader improves by over 50% relative to the closest system from previous work in predicting grades supplied by human judges.