基于多因素语义相似度的主观题自动评阅方法研究  

Study of automatic marking method of subjective questions based on semantic similarity

作　　者：党丽琼[1,2] 刘文辉[3] 

机构地区：[1]陕西航天职工大学 [2]西安石油大学 [3]95092部队

出　　处：《计算机时代》2012年第5期9-12,共4页Computer Era

摘　　要：主观试题自动评阅系统涉及人工智能、模式识别以及计算机自然语言理解等复杂的系统知识,是无纸化考试中自动阅卷系统的设计难点。为此,仿照传统考试过程中人工阅卷的基本思路,把主观试题的自动评阅问题分解为知识库的构建、概念相似度的计算以及语句结构的处理等三部分,并把一种改进的多因素语义相似度算法引入到概念相似度计算,详细设计了主观试题的计算机自动评阅模型。实例运行结果表明,占总数70%的计算结果在可接受误差范围内。It is because of the automatic marking of the subjective questions involving artificial intelligence,pattern recognition and computer understanding of natural language that has become a difficulty of the automatic marking in the design of paperless examination.By simulating the basic train of thought of the manual marking in traditional test,automatic marking of the subjective questions is divided into three parts which are construction of knowledge,calculation of concept similarity and handle of sentence of structure,then a kind of improved multi-factor semantic similarity calculation is introduced to solve the concept similarity,finally a computer automatic marking model of subjective questions is designed in detail.Practical operation results show that seventy percent of the total results within the acceptable error range.

关 键 词：主观试题 多因素 自动评阅 语义相似度 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]