【摘要】 随着Internet上信息量的飞速增长,成千上万的网上文档需要分类以方便用户的浏览和获取。因此文档的自动分类工作已经越来越受到重视,一些相应的分类方法也应运而生。但其中很少有涉及到"层次化"的分类领域,且绝大多数方法仅仅返回单个分类结果。文中,我们提出了一种新的文档自动分类方法:MRHC(Multicategory ReturnedAlgorithmforHierarchicalClassification)。该方法着眼于层次化的分类技术,并在适当的情况下为文档返回多个分类结果。该方法中结合了特征削减和增量学习技术以便提高分类性能。最后,为了更加准确、客观的评价分类结果,提出了一种新的评估方法:LEP(Length of Error Path)。实验结果表明,提出的分类方法响应时间短,分类准确度高,具有较强的实用性。更多还原

【Abstract】 Automatic text classification has become more and more important as a result of information explosion on the Internet. Thousands of documents need to be classified so that users can access them easily. Numerous approaches have been proposed to solve such a problem, but few of them deal with the hierarchical classification case, and most of them only return one category result. In this paper, we present a straightforward algorithm MRHC, which aims at hierarchical classification and assigns the testing document to multiple categories if necessary. Feature reduction and incremental training technology are also used to improve the classifying performance. A new evaluation metric based on the LengthofErrorPath (LEP) is also proposed to evaluate the accuracy of hierarchical classification more objectively. Experiments demonstrate the feasibility of our approach with short system responding time and high accuracy.更多还原