NTCIR-5 WEB Navi-2 Experiments at Osaka Kyoiku University - Page, Anchor and Title Indexing, and In-link Count, Inter Page and Inter Site Link Analyses
NTCIR-5 WEB Navi-2 Experiments at Osaka Kyoiku University - Page, Anchor and Title Indexing, and In-link Count, Inter Page and Inter Site Link Analyses
复制标题
大阪教育大学的 NTCIR-5 WEB Navi-2 实验 - 页面、锚点和标题索引以及内链接计数、页面间和站点间链接分析
DOI：
--
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
H. Nakakubo
中科院分区：
文献类型：
--
作者：
Takashi Sato;H. Nakakubo
本文描述了WEB导航检索子任务2 （WEB Navi-2）的实验结果。我们制作了三个基于克的索引，即全页文本索引、标题标签中的文本索引和锚标签中的文本索引。由于基于gram的索引能够索引目标文本中的所有字符串，因此在字典中找不到的单词基本上也会被索引。我们使用搜索主题TITLE标签中的单词作为查询。我们做了三种类型的链接分析，即内链接数和站点间和页面间链接分析。我们合并了三个指标的词搜索得分和不同的链接分析得分。我们发现锚文本分析对于WEB Navi-2是最有效的，并且有必要设计页面和/或标题分数与锚分数的合并。
This paper describes experimental results of WEB Navigational Retrieval Subtask 2 (WEB Navi-2). We made three gram-based indices, namely indices for text in whole page, text in title tag and text in anchor tag. Since gram-based indices are able to index all strings in target text, words that are not found in dictionaries are also indexed essentially. We used words in TITLE tag of search topics as queries. We did three kinds of link analyses, that is, in-link count and inter site and inter page link analysis. We merged score from word search for three indices and score from link analyses variously. We found that anchor text analysis was most effective for WEB Navi-2, and that it is necessary to devise merging of page and/or title score to anchor score.