摘要
为了对突发事件Web新闻进行更精确的分类,研究了突发事件Web新闻的多层次自动分类方法.该方法初步分析了突发事件Web新闻的分类,给出3层分类器的构造方法,即第1级和第2级通过规则定制来完成,第3级通过统计学习训练并实现,并研究了HTML文本向量空间模型及特征项的抽取方法.将该自动分类方法在甲型H1N1、法国空难以及汶川大地震等突发事件的Web新闻中进行了训练和测试.实验结果表明,所提方法的分类效果优于改进前的方法.
To accurately classify Web news of unexpected events,an automated multiple hierarchical classification method is proposed.Firstly,the classification of Web news of unexpected events is analyzed in brief and three classifiers are constructed.The first and the second classifier are accomplished by the rules,while the third one needs to be realized by machine learning.Much more attention then is paid to Vector Space Model and feature extraction in HTML text.At last,the experiments which include the subjects of H1N1 influenza,French air crash and Wenchuan earthquake are conducted and analyzed.The results show that the method introduced is better than traditional methods.
出处
《北京工业大学学报》
EI
CAS
CSCD
北大核心
2011年第6期947-954,共8页
Journal of Beijing University of Technology
基金
国家自然科学基金资助项目(70971005
90924020)
国家科技支撑计划重大专项(2006BAK04A23)
质检公益行业科研专项(200910088)
关键词
文本分类
分类器
特征抽取
多层次体系
突发事件
text classification
classifiers
feature extraction
hierarchical systems
unexpected events