Research on Question Severity Detection for Automatic Essay Scoring

Abstract

In the process of essay scoring, the item-specific index is an important factor affecting the final score. In previous studies, the item-specific index was not included in the characteristic system of objective essay scoring. Therefore, it is easy to produce the phenomenon that the deviation from the topic with high language standardization, rich vocabulary and correct grammar is judged high, resulting in the decrease of the accuracy of the score prediction model, which is quite different from the results of manual evaluation. In this paper, the LDA topic model is used to model the article, and the topic probability distribution distance and word vector are used to calculate the topic degree index of the article. The index is added to the composition automatic scoring system to improve the accuracy of the machine scoring.