The Intelligent Composition Evaluation System Based on Anomaly Detection

Abstract

Compositions evaluation is one of the main tasks in Chinese teaching, how to evaluate compositions quickly and efficiently is an important assignment of language intelligence research. A method combining artificial intelligence and language intelligence is proposed to solve this problem, constructing a compositions evaluation system based on anomaly detection which belongs to unsupervised learning algorithm, applying Gaussian distribution theory to distinguish these abnormal characteristics from normal ones in a batch of compositions, according to which method could quickly identify anomalies in the compositions lead to low scores and ultimately achieve the goal of evaluating compositions accurately and efficiently.