Design and Evaluation of Maintenance Tools for Distributed Digital Libraries

这项有针对性的研究项目是研究和开发工具和社会协议，使分布式数字图书馆馆藏的管理和维护变得更加可行。在分布式数字图书馆馆藏中，作者将材料放入图书馆，图书馆员(馆藏管理员)为图书馆读者组织和注释。虽然这种“基于作者”的方法使数字图书馆能够迅速发展，但它们可能会造成混乱，因为资源在没有警告的情况下被添加、删除或更改。在指向网站的情况下，注意这些网站何时消失、重新托管或更改其底层结构是非常耗时的。此外，当对资源的内容进行更改时，集合管理器必须决定新版本是否仍然适合集合，如果是，则决定文档是否需要重新分类。为了提高馆藏管理员维护分布式数字图书馆的能力，该项目正在开发：i)识别不再可用的资源的算法和启发式方法；ii)识别资源重新定位的方法；iii)对资源变化的重要性进行分类和评估的方法；以及iv)支持社会机制的工具(在资源作者、图书馆管理者和图书馆读者之间)，以应对文件的变化。该项目计划包括根据技术和社会指标对识别和分类算法进行评估。这些评估回答了算法是否正确识别网络和服务器问题，是否成功定位了已移动的资源，以及更改的重要性评级是否与人工评估者的评级相匹配。

This targeted research project is investigating and developing tools and social protocols to make more feasible the management and maintenance of distributed digital library collections in which authors put material into the library and librarians (collection managers) organize and annotate it for the library patrons. While such "author-based" approaches enable a digital library to grow rapidly, they can create confusion as resources are added, deleted, or changed without warning. In the case of web sites that are pointed to, noticing when those web sites go away, are rehosted, or change their underlying structure is very time consuming. Additionally, when changes are made to the content of the resources, the collection manager must decide if the new version is still suitable for the collection and, if so, whether the document needs to be recategorized. Towards the goal of improving the ability of collection managers to maintain distributed digital libraries this project is developing: i) algorithms and heuristics for identifying resources no longer available; ii) methods for identifying the relocation of resources; iii) methods for categorizing and evaluating the significance of changes to resources; and iv) tools supporting social mechanisms (between resource authors, library managers, and library patrons) to contend with document changes. The project plan includes the evaluation of identification and categorization algorithms based on technical and social metrics. These evaluations answer whether the algorithms correctly identify network and server problems, whether resources that have been moved are successfully located, and whether ratings of significance of change match those of human evaluators.