| 本文已被：浏览 5557次 下载 8480次 |
码上扫一扫！ |
|
|
| 基于网页格式信息量的博客文章和评论抽取模型 |
|
曹冬林1,2,3, 廖祥文1,2, 许洪波1, 白硕1
|
|
1.中国科学院 计算技术研究所 网络科学与技术研究部,北京 100190;2.中国科学院 研究生院,北京 100049;3.厦门大学 智能科学系,福建 厦门 361005
|
|
| 摘要: |
| 从信息论的角度出发,提出了一个基于网页格式信息量的博客文章和评论抽取模型.首先,结合网页视觉上的位置信息和文本的有效信息来定位网页正文.其次,利用博客网页中的格式信息作为信息单元并计算每个信息块所包含的格式信息量,通过计算最小切分位置信息量来切分正文中的文章和评论.该模型具有与语言无关的特点,因此具有一定的通用性.实验结果表明,该模型在博客正文定位和正文切分方面达到了较高的精确率. |
| 关键词: 博客信息抽取 最小正文子树 有效信息率 网页格式信息 视觉信息 切分位置信息量 |
| DOI： |
| 分类号: |
| 基金项目:Supported by the National Basic Research Program of China under Grant Nos.2004CB318109, 2007CB311100 (国家重点基础研究发展计划(973)); the National High-Tech Research and Development Plan of China under Grant No.2007AA01Z441 (国家高技术研究发展计划(863)) |
|
| Extraction Model Based on Web Format Information Quantity in Blog Post and Comment Extraction |
|
CAO Dong-Lin,LIAO Xiang-Wen,XU Hong-Bo,BAI Shuo
|
| Abstract: |
| Based on the information theory, this paper presents a model based on Web format information quantity in blog information extraction. First, the vision information in blog Web page and the effective text information are combined to locate the main text which represents the theme of the blog Web page. Second, the format information of blog Web page is used to calculate the information quantity of each block and the minimal separating information quantity of separate position is used to detect the boundary of posts and comments in the main text. This model is language insensitive and can be used in a lot of blogs which are written in different natural languages. Experimental results show that this method achieves high precision in locating main text and separating the post and comment. |
| Key words: blog information extraction minimal main text subtree effective information ratio Web format information vision information information quantity of separate position |
