【作者】 骆昌日；

【导师】 何婷婷；

【作者基本信息】 华中师范大学 ， 计算机应用技术， 2004， 硕士

【摘要】 随着信息技术的发展，人们已经从信息缺乏的时代过渡到信息极为丰富的数字化的时代。如何从这些海量信息中迅速有效地获得所需信息也就成为一项很重要的研究课题。为此目的，文本自动分类被研究者提出并进行了应用研究。研究文本自动分类具有重要意义，它可以大大的缩短了对资料的整理时间，为信息检索提供方便，有利于现实文档的存档管理。 本文主要是应用统计的方法对文本自动分类进行理论和实践的探讨。我们的工作主要从以下方面进行： 1．探讨了基于统计方法的文本自动分类的定义、常用模型和常用算法。 2．讨论了文本自动分类器一般方法、步骤及有关技术细节。 3．在向量空间模型下，实现了向量距离加权算法、代表向量算法、中心向量算法构造的三种分类器。对三种分类器分别以字、词为特征进行分类测试、分析发现：①使用相同的分类算法，用词作为特征项，比以字作为特征的分类效果好；②用不同的算法构造分类器对分类效果的影响很大，如中心向量算法在字、词特征下的分类效果优于其他两算法；在以字为特征的情况下，该算法的平均查全率80.73％，平均查准率82.94％；在以词为特征的情况下，该算法的平均查全率83.6％，平均查准率85.97％；③选用语料不同对分类效果也有影响，如用新浪网(www.sina.com.cn)网页语料进行测试，使用中心向量法分类器和词作为特征的情况下，平均准确率为89.31％，平均查全率为88.33％。 4．基于改进后的中心向量法重构自动分类器，测试取得了开放测试平均查全率90.35％、平均准确率90.87％和封闭测试平均查全率98.36％、平均准确率98.74％的分类效果，说明改进后的算法适合中文文本分类。 本文所得到的这些实验数据对于开发实际的文本分类系统具有指导意义。该研究可应用于网络信息检索、信息过滤、中文文本自动分类、中文网页自动分类等应用领域。更多还原

【Abstract】 With the development of information technique, people have already transited into the ages in which information is extremely abundant and digitized from the age lacks information. How to acquire the useful information quickly and effectively from information-sea has become a very important problem. For this purpose, the text automatic classification has been put forward and studied in application.This paper gives details to the research on the theory and practice of the text automatic classification using the statistical method. The main aspects of the paper are as follows:1. The definition, common used models and common used algorithms of classification are discussed theoretically.2. Discuss the general methods and the key technology of constructing classifier.3. We employ vector-distance weighted algorithm, representative-vector-dista -nee algorithm and center-vector algorithm to construct the classifier. And then, the experiments of the three classification algorithms have been done respectively with different feature-set (Chinese-character feature-set and Chinese-word feature-set). According to the analysis of the experimental results, we find that: Qthe classification result with the same classifier by taking Chinese-word as feature is better than by Chinese-character. (2)the influence to classification result is highly effected by using different classifier, for example, the center-vector algorithm obtains better classification results than other two algorithms. With the character feature, the average recall is 80.73%, and the average precision is 82.94%, and with the Chinese-word feature, the average recall is 83.6%, and the average precision is 85.97%.Different corpuses influence the classification result. For example, the average recall is 89.31% and the average precision is 88.33%, by using the news web pages as corpus from the web site "www.sina.com.cn", which adopt the center-vector algorithm to structure classifier and select Chinese-word as feature.4. For the improved algorithm experimental results, the average recall is 96.35%, and the average precision is 90.87%. The experimental results indicate that the improved algorithm is suit for Chinese text automatic classification.This study can be used in network information retrieve, information filter, Chinese text automatic classification, Chinese web page automatic classification and other application fields.更多还原