基于相似度的网页标题抽取方法_李国华精选.pdfVIP

25 2 V ol . 25, No . 2 20 11 3 J OU RNA L OF CH INESE IN FO RM AT ION PR OCESSIN G M ar., 2011 : 2011) 0200 3206 , , 450001) : 目前网页标题的抽取方法大多结合 H T M L 结构和标签特征进行抽取, 但是这些方法并没有考虑标题 正文信息之间内容上的联系该文提出一种基于相似度的网页标题抽取方法, 该方法利用网页标题 正文信息之 间的关系, 通过计算语言 单位之间的相似度和对应的权值, 并引入 HIT S 算法模型对权值进行调整, 根据特定的 选取方法抽取出真实标题实验结果表明, 该方法不仅对 非标准网页的抽取达到满意的效果, 而且对 标准网 页具有较高的泛化能力 : 网页标题抽取; 相似度; W eb 信息抽取 : T P39 1 : A Title Extraction from HTML Documents Based on Similarity LI Guohua, ZAN Hongying Co llege of Info rmation Eng ineer ing, Zheng zho u U niv ersity , Zhengzhou , H enan 450001, China) Abstract: Most of the methods for title extraction from HT ML documents are based on the structure of HT M L document or the f eatures of label. T hey do not considered the co rrelation betw een the title and the content . T his paper proposes a method of title ex traction from HT ML documents based on similarity, w hich makes full use of the correlation betw een the title and the main body. T he similarity betw een units are calculated and adjusted by the HIT S algorithm . Then the real title is extracted in a ser ies of steps. Ex perimental results show that this method perform s w ell for nonstandard HT M L document and has good generalization ability for standard HT M L document. Key words: t itle ex tr act ion ; similar ity ; Web infor matio n r etriev al ; 1 ; H T M L , , , ) , H T ML t it le / tit le , , , H T M L