A Human-Computer Collaboration Approach to Improve Accuracy of an Automated English Scoring System

This paper explores an issue of redundant errors reported while automatically scoring English learners’ sentences. We use a human-computer collaboration approach to eliminate redundant errors. The first step is to automatically select candidate redundant errors using PMI and RFC. Since those errors are detected with different IDs although they represent the same error, the candidacy cannot be confirmed automatically. The errors are then handed over to human experts to determine the candidacy. The final candidates are provided to the system and trained with a decision tree. With those redundant errors eliminated, the system accuracy has been improved.