• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Information Processing & Management >Towards a user-oriented thesaurus for non-domain-specific image collections
【24h】

Towards a user-oriented thesaurus for non-domain-specific image collections

机译：面向面向用户的同义词库，用于非特定域的图像集合

获取原文
获取原文并翻译 | 示例
       
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

This study explored how user-supplied tags can be applied to designing a thesaurus that reflects the unique features of image documents. Tags from the popular image-sharing Web site Flickr were examined in terms of two central components of a thesaurus-selected concepts and their semantic relations—as well as the features of image documents. Shatford's facet category and Rosch et al.'s basic-level theory were adopted for examining concepts to be included in a thesaurus. The results suggested that the best approach to Color and Generic category descriptors is to focus on basic-level terms and to include frequently used superordinate- and subordinate-level terms. In the Abstract category, it was difficult to specify a set of abstract terms that can be used consistently and dominantly, so it was suggested to enhance browsability using hierarchical and associative relations. Study results also indicate a need for greater inclusion of Specific category terms, which were shown to be an important tool in establishing related tags. Regarding semantic relations, the study indicated that in the identification of related terms, it is important that descriptors not be limited only to the category in which a main entry belongs but broadened to include terms from other categories as well. Although future studies are needed to ensure the effectiveness of this user-oriented approach, this study yielded promising results, demonstrating that user-supplied tags can be a helpful tool in selecting concepts to be included in a thesaurus and in identifying semantic relations among the selected concepts. It is hoped that the results of this study will provide a practical guideline for designing a thesaurus for image documents that takes into account both the unique features of these documents and the unique information-seeking behaviors of general users.
机译：这项研究探索了如何将用户提供的标签应用于设计反映图像文档独特功能的同义词库。我们从词库选择的概念的两个主要组成部分及其语义关系以及图像文档的功能方面对来自流行的图像共享网站Flickr的标签进行了检查。运用Shatford的构面类别和Rosch等人的基本层次理论来检查要包含在同义词库中的概念。结果表明，颜色和通用类别描述符的最佳方法是专注于基本级别的术语，并包括常用的上级和下级术语。在“抽象”类别中，很难指定一组可以一致且占优势的抽象术语，因此建议使用层次关系和关联关系来增强可浏览性。研究结果还表明，需要更多地包含特定类别的术语，这些术语被证明是建立相关标签的重要工具。关于语义关系，该研究表明，在相关术语的识别中，重要的是，描述符不仅限于主要条目所属的类别，还应扩大其范围以包括其他类别的术语。尽管需要进行进一步的研究以确保这种面向用户的方法的有效性，但这项研究仍取得了可喜的结果，表明用户提供的标签可以在选择要包含在同义词库中的概念以及识别选定对象之间的语义关系方面提供帮助。概念。希望这项研究的结果将为设计图像文档同义词库提供实用的指导，同时考虑这些文档的独特功能和普通用户的独特信息搜寻行为。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号