权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

A Fusion Method of Web Archives and Video Archives for Next Generation Digital Libraries

下一代数字图书馆网络档案与视频档案的融合方法

基本信息

批准号：
16300028
负责人：
SUMIYA Kazutoshi
金额：
$ 9.22万
依托单位：
University of Hyogo
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (B)
财政年份：
2004
资助国家：
日本
起止时间：
2004 至 2006
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-16300028/
关键词：
Metadata Video content Web content Web archives Temporal Relation Interpolation Video archives Gaze information 補完オンライン地図

项目摘要

In this research, we discussed the managing, storing and retrieving method for large-scale Web archives which consist whole Web contents from past to now. The aim of this research is to achieve the digital contents management infrastructure in the digital TV broadcast systems and the broadband Internet.We have developed a search method that is sensitive to the temporal relation between query keywords, allowing use of the relation to retrieve a time series of Web pages from Web archives. That is, retrieved URLs can be categorized based on the temporal relation. Web pages are extracted by generating queries which extract time intervals that satisfy those conditions. Once a temporal relation is determined, it can be expanded and used for related URLs, thus enabling certain Web archive search engine functionalities.Secondly, we have proposed a framework for integrating news video streams and blogs. In addition, we defined the evaluation structure, which consists of the object word and the … More evaluation ratio. The method of using the evaluation structure that we proposed has two parts: a) calculating news evaluation based on scene order, and b) searching blogs using the extracted news evaluation. In the experiment, we showed that the scene order of news video stream could be reacted on users' impression as well as Kuleshov effect in films. In addition, blogs extracted by the prototype included not only information on each news topic, but also evaluations of the news topics.Finally, we developed a retrieval system for displaying appropriate Web pages by defining and analyzing user operations with map interfaces as chunk. We experimentally found that our method was effective for retrieving individual points and confirming routes. The Web pages retrieved by our proposed method during the experiments were considered appropriate. However, some pages were considered to be not relevant to the users' intentions. We need to evaluate the method using further various and practical data and must improve the definition of the operation sequences. Less

在本研究中，我们讨论了从过去到现在对构成整个Web内容的大规模Web档案的管理、存储和检索方法。为了实现数字电视广播系统和宽带互联网中的数字内容管理基础设施，我们开发了一种对查询关键字之间的时间关系敏感的搜索方法，允许利用这种关系从Web档案中检索Web页面的时间序列。也就是说，可以基于时间关系对检索到的URL进行分类。通过生成提取满足这些条件的时间间隔的查询来提取网页。一旦确定了时间关系，就可以将其扩展并用于相关的URL，从而实现特定的Web档案搜索引擎功能。其次，我们提出了一个整合新闻视频流和博客的框架。此外，我们还定义了由对象词和…组成的评价结构更多的评价比率。使用我们提出的评价结构的方法包括两个部分：a)基于场景顺序计算新闻评价；b)使用提取的新闻评价来搜索博客。在实验中，我们发现新闻视频流的场景顺序不仅可以反映电影中的库拉舍夫效应，还可以反映用户的印象。此外，原型提取的博客不仅包含每个新闻主题的信息，还包含对新闻主题的评价；最后，通过定义和分析以地图界面为块的用户操作，开发了一个显示合适Web页面的检索系统。我们通过实验发现，我们的方法对于检索单个点和确定路径是有效的。我们提出的方法在实验期间检索到的网页被认为是合适的。然而，一些页面被认为与用户的意图无关。我们需要进一步使用各种实际数据对该方法进行评估，并必须改进操作顺序的清晰度。较少