Integrating various features to grade students' writings based on improved multivariate Bernoulli model

Most of AES systems use various writing features as the scoring criteria. A valid mathematical model then integrates the various features demonstrated in the essay and calculates a possible score. Many studies have addressed the advantages and necessity of using multiple features to predict the score of an essay. These traditional prediction models have difficulty integrating various types of numerical data. However, the values, calculated by the features proposed by recent researches, are often not continuous numerical values. This study will propose an improved multivariate Bernoulli model to integrate features to predict an essay score even the value types of features are different