基于文档发散度的作文跑题检测Off-topicEssaysDetectionBasedon.PDF

文章编号： 基于文档发散度的作文跑题检测 1,2 1,2 陈志鹏 ，陈文亮 （1.苏州大学 计算机科学与技术学院，江苏 苏州 215006 ；2.软件新技术与产业化协同创新中心，江苏 苏 州 215006 ） 摘要：作文跑题检测是作文自动评分系统的重要模块。传统的作文跑题检测一般计算文章内容相关性作为 得分，并将其与某一固定阈值进行对比，从而判断文章是否跑题。但是实际上文章得分高低与题目有直接 关系，发散性题目和非发散性题目的文章得分有明显差异，所以很难用一个固定阈值来判断所有文章。本 文提出一种作文跑题检测方法，基于文档发散度的作文跑题检测方法。该方法的创新之处在于研究文章集 合发散度的概念，建立发散度与跑题阈值的关系模型，对于不同的题目动态选取不同的跑题阈值。本文构 建了一套跑题检测系统，并在一个真实的数据集中进行测试。实验结果表明基于文档发散度的作文跑题检 测系统能有效识别跑题作文。 关键词：跑题检测；文档发散度；文本相似度 中图分类号：TP391 文献标识码：A Off-topic Essays Detection Based on Document divergence CHEN Zhipeng1,2, CHEN Wenliang1,2 (1.School of Computer Science and Technology, Soochow University ,Suzhou ,Jiangsu 215006, China; 2.Collaborative Innovation Center of Novel Software Technology and Industrialization, Suzhou ,Jiangsu 215006, China) Abstract: Off-topic detection is important in the automated essay scoring systems. Traditional methods compute similarity between essays and then compare the similarity with a fixed threshold to tell whether the essay is off-topic. In fact, there is strong relationship between essay score and the type of topic. It is obviously different between divergent topic and non-divergent topic. It is hard to use a fixed threshold to identify off-topic for all essays. This paper proposes a new method of off-topic detection based on divergence of essays. The innovation of this paper is that we study the divergence of essays, and establish the linear regression model between divergence and threshold. In our method, we set a threshold for each topic dynamically. Finally we establish an off-topic detection system, and test in the real data. Experimental results show that our method is more effective than baseline systems. Key words: off-topic detection; document divergence; document similarity 1 引言 作文跑题指文章偏离