Despite the plethora of born-digital content, vast troves of important content remain accessible only on physical media such as paper or microfilm. The traditional approach to indexing undigitized content is using manually created metadata that describes content at some level of aggregation (e.g., folder, box, or collection). Searchers led in this way to some subset of the content often must then manually examine substantial quantities of physical media to find what they are looking for. This paper proposes a complementary approach, in which selective digitization of a small portion of the content is used as a basis for proximity-based indexing as a way of bringing the user closer to the specific content for which they are looking. Experiments with 35 boxes of partially digitized US State Department records indicate that box-level indexes built in this way can provide a useful basis for search.

翻译：尽管数字原生内容浩如烟海，仍有大量重要资料仅能以纸张或微缩胶片等物理媒介形式获取。传统上，对未数字化内容进行索引需依靠人工创建的元数据，这些元数据按某种聚合层级（如文件夹、档案盒或全集）描述内容。通过这种方式被引导至部分内容的检索者，往往还需手动查阅大量物理媒介才能找到目标。本文提出一种补充方法：通过对一小部分内容进行选择性数字化，以此为基础构建近邻索引，从而将用户更精准地引向其所需的具体内容。基于美国国务院35箱部分数字化档案的实验表明，以这种方式构建的档案盒级索引可为搜索提供有效基础。