Page 168 - Journal of Library Science in China, Vol.47, 2021

NIU Li, GAO Chenxiang, ZHANG Yufeng, YAN Shi, XU Yongjun & LI Anrunze / Discovering, reorganizing 167
and storytelling: paths and methods of archives research on the perspective of digital humanities

Figure 5. Technical system of archival data research from DH perspective

3.1 Archival data processing technology from the perspective of value preservation

Archival data processing technology from the perspective of value preservation is used to support
the “discovering” session in archival data research methodology. In the extraction of archival data
elements, this session is divided into three main parts: “metadata annotation”, “object detection
and extraction” and “context recognition”. At the level of “metadata annotation”, consideration
should be given to embedding new electronic record metadata standards such as “signature”,
“confirmation” and “format permanence” in the traditional archival metadata classification
[33]
system , so that when digital archives are transformed into archival data from DH perspective,
their source relationships and evidentiary characteristics are retained in the process of granular
segmentation and entity reorganization, and they are treated as attributes of relevant archival
resources.
The “Object Detection and Extraction” session focuses on the application of natural language
processing and image recognition framework based on deep learning, extracting key entities such
as characters, buildings, and time from standardized texts, images, and videos, and using objective
algorithms combined with the form of manual recognition to extract entities and their relationships
in archival resources, so as to supplement the semantic connections between entities on the basis of
avoiding subjective bias and objective technical defects.
“Context Recognition” emphasizes the identification and association of elements such as
archival creators, users, element structures, archival functions and activities, business scenarios
and institutional functions. One of the essential differences between archival resources and other
information resources is retaining context [34] . At present, the context identification process of
archival resources still relies on the basic methods and skills of humanities research, which requires