多分类器融合技术在自动作文评分中的应用

The Application of Classifier Combining in Automated Essay Scoring

从作文的内容和语言学两个方面抽取了作文中相关的特征,并利用多种分类器(贝叶斯、K近邻和支持向量机)根据各方面的特征实现了对作文的分类(评分).最后利用多分类器融合技术对多个分类器进行了融合处理.通过实验分析,利用文本分类的方法对作文进行评分是完全可行的,在采用融合技术以后的评分性能有了较大的提高.

We aim to abstract related features from an essay by analyzing its content and structures. Meanwhile, classifiers including Bayes, KNN and SVM are adopted to realize better classification of essays （scoring） based on their features from various aspects. The multi-combination technology is also used in combining different classifiers. Through experimental analysis, it is indicated that scoring for essays is highly feasible via text-classifying method, and a higher performance is obtained by adopting multi-classifiers technology than previously single-classifier one.