| 基于优选链接的中文网页分类方法研究 |
| |
| 引用本文： | 陈胜荣,董守斌.基于优选链接的中文网页分类方法研究[J].郑州大学学报(理学版),2007,39(2):78-82. |
| |
| 作者姓名： | 陈胜荣 董守斌 |
| |
| 作者单位： | 华南理工大学计算机科学与工程学院,广州,510640 |
| |
| 摘 要： | 网页之间的链接为Web数据挖掘提供了丰富信息,通过链接关系来加强中文网页分类的效果.由于网页的编写存在随意性和不规范性,并非所有的链接页面与源页面都有主题相关性.通过比较页面与其链接页面的相似性,优选出相似度较高的链接子集,利用得分向量对基于内容的分类结果进行修正.基于北大天网数据集的实验结果表明,优选链接页面的分类方法对于训练样本较少的类别在分类精度上有一定的提高.
|
| 关 键 词： | 分类 链接 相似度 得分向量 |
| 文章编号： | 1671-6841(2007)02-0078-05 |
| 修稿时间： | 12 8 2006 12:00AM |
Research on Chinese Web Pages Classification Based on Preferential Links |
| |
| CHEN Sheng-rong,DONG Shou-bin.Research on Chinese Web Pages Classification Based on Preferential Links[J].Journal of Zhengzhou University:Natural Science Edition,2007,39(2):78-82. |
| |
| Authors: | CHEN Sheng-rong DONG Shou-bin |
| |
| Institution: | College of Computer Science and Engineering, South China University of Technology, Guangzhou 510640, China |
| |
| Abstract: | Links among pages provide rich information for Web mining.Links relationship improve the Chinese Web pages classification.As Web pages are created casually,not all linked pages share the same topic as the source page.So,by comparising the similarity between page and its linked pages,a subset of the linked pages that has the highest similar score is chosen and the original classification results are modified by score vector.The experiment based on the Chinese dataset provided by Tian Wang of Peking University shows that this method improves classification in recall,precision as well as F1 value. |
| |
| Keywords: | classification links similarity score vector |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
