Automated Essay Scoring Using Multi-classifier Fusion

The method of multi-classifier fusion was applied to essay scoring. In this paper, each essay was represented by Vector Space Model (VSM). After removing the stopwords, we extracted the features of contents and linguistics from the essays, and each vector was expressed by corresponding weight. Three classical approaches including Document Frequency (DF), Information Gain (IG) and Chi-square Statistic (CHI) were used to select features by some predetermined thresholds. According to the experimental results, we classified the test essay to appropriate category using different classifiers, such as Naive Bayes (NB), K Nearest Neighbors (KNN) and Support Vector Machine (SVM). Finally the ensemble classifier was combined by those component classifiers. After training for multi-classifier fusion technique, the experiments on wCET4 essays about same topic in Chinese Learner English Corpus (CLEC) show that precision over 73% was achieved.