摘要
加减文字题是小学数学的一个重点和难点问题,长期以来,人们对加减文字题的研究局限于教育学和心理学领域,该文从文本分类以及信息检索的角度出发,将加减文字题中的算术关系看成是一个分类问题,尝试用机器学习的方法来对其进行分类,分别研究了人工神经网络（ANN）和支持向量机（SVM）在加减文字题算术关系抽取中的应用,并对两种算法的试验结果进行了比较和分析。通过分词,关键词选取,构造特征向量,分别运用两种算法对其进行分类。对试验结果进行评测发现,在一定条件下SVM算法明显优于ANN算法。
Addition and subtraction word problems are found very difficult to both teacher and children in elementary school. The research on addition and subtraction word problems has been confined to education and psychology fields for a long time. In a viewpoint of text categorization and information retrieval, the arithmetical relation in these problems is considered as a categorization task and studied with machine learning methods. In this paper, word problems are segmented firstly into words by software. Some
作者
苏林忠
SU Lin-zhong （The Office of Information, South China University of Technology, Guangzhou 510640, China）
出处
《电脑知识与技术》
2010年第7期5302-5304,共3页
Computer Knowledge and Technology
关键词
加减文字题
人工神经网络
支持向量机
关系抽取
文本分类
中文信息处理
are selected
and formed into feature vector. Two machine learning methods
Support Vector Machines （SVM） and Artificial Neural Network （ANN）
are compared when they are used in the relation extraction respectively. The empirical result indicates that SVM is more effective than ANN in our experiments. Key words： add and subtraction word problems
ANN
SVM
relation extraction
text categorization
Chinese information processing
