摘要
现有的文本自动分类离不开文档向量的构造,向量的分量与文档中的特征项相对应。这种向量通常高达几千维甚至数万维,计算量相当大,因此需要对向量进行约简。而传统的基于频率的阈值过滤法往往会导致有效信息的丢失,影响分类的准确度。该文将Rough集理论引入自动分类,并提出了一种新的文档向量约简算法。实验证明该算法不仅能有效缩减文档向量的规模,而且相比传统的阈值法信息损失小、准确率更高。
Much of the previous automatic Text Classification (TC) methods are closely connected with the construction of document vectors. With each term corresponding to a unit in the vector, this method maps the document vectors into a very high dimensional space, possibly of tens of thousands of dimension, which results in a massive amount of calculation. Since the traditional algorithms based on frequency and threshold filtering may often lead to the loss of effective information, this paper presents a new system for TC, which introduces rough set theory that can greatly reduce the document vector dimensions by reduction algorithm. The empirical results prove to be very successful, for it can not only effectively reduce the dimensional space, but also reach higher accuracy while losing less information compared with usual reduction methods.
出处
《电子与信息学报》
EI
CSCD
北大核心
2005年第7期1047-1052,共6页
Journal of Electronics & Information Technology
基金
教育部优秀青年教师资助计划教育部归国人员启动基金模式识别国家重点实验室开放基金清华大学基础研究基金资助课题
关键词
自动分类
ROUGH集
决策表
约简算法
Automatic classification, Rough set, Decision table, Reduction algorithm
