Visual Analytics for the image archive of the German Colonial Society (VaBiKo)

位于法兰克福的约翰·克里斯蒂安·森肯伯格大学图书馆的德国殖民学会图像档案馆收藏了德国殖民机构最重要的照片和出版物之一，约有55,000张图像载体。然而，目前，德国殖民协会图像集的数字化版本不符合当前的质量标准，图像的可检索性因元数据不完整而受到限制。这严重限制了研究人员对图像集合的使用。拟议的项目解决了这些关切，旨在显著改善德国殖民学会图像档案对研究人员的获取和使用情况。增加对学术内容的获取将为继续重新评估德国殖民史奠定基础。通过对图像内容进行全面的半自动索引，文件将大大扩展。在这一过程中产生的额外元数据将有助于提高图像资料的可搜索性。该项目将可持续地提高图像收集的可见性和可获得性。此外，该项目还将解决与藏品使用有关的伦理问题。现有界面将发展成为一个多语种门户，用于图像(和文本)持有物的交互式语义搜索。研究进程将通过以下功能得到支持：-增强探索和展示选择；-利用深度学习技术改进图像保有量和历史图像来源索引的可搜索性；-空间和时间信息的交互式可视化以及基于相似性的图像分组和历史幻灯片图像与文本的链接；-根据当前标准加强数据管理，包括改进的互操作性(界面和应用编程接口)。

The image archive of the German Colonial Society at the Johann Christian Senckenberg University Library in Frankfurt houses one of the most important collections of photographs and publications of German colonial institutions, comprising approximately 55,000 image carriers. However, at the moment, the digitized versions of the German Colonial Society image collection are not available in current quality standards and the retrievability of the images is limited by incomplete metadata. This severely limits researchers’ usage of the image collection. The proposed project addresses these concerns and aims to significantly improve access and usability of the German Colonial Society's image archive for researchers. Enhanced access to content for scholarship will lay the foundation for the continued reappraisal of German colonial history. The documentation will be significantly expanded through comprehensive semi-automatic indexing of the image content. The additional metadata generated in the process will serve to improve the searchability of the image holdings. The project will sustainably increase the visibility and findability of the image collection. In addition, the project will address ethical issues related to the use of the collection. The existing interface will be developed into a multilingual portal for interactive, semantic search in image (and text) holdings. The research process will be supported through the following functionalities: - Enhanced exploration and presentation options; - Improved searchability of image holdings and indexing of historical image sources using deep learning techniques; - Interactive visualizations of spatial and temporal information as well as similarity-based grouping of images and the linking of images and text for the historical slideshows; - Enhanced data management according to current standards including improved interoperability (interfaces and APIs).