Page 164 - Journal of Library Science in China, Vol.47, 2021
P. 164

NIU Li, GAO Chenxiang, ZHANG Yufeng, YAN Shi, XU Yongjun & LI Anrunze / Discovering, reorganizing   163
                                        and storytelling: paths and methods of archives research on the perspective of digital humanities

               structure or different structures to form a knowledge network of a specific theme [27] . However,
               the two-dimensional network knowledge organization form cannot realize a multi-dimensional
               mutual verification of basic attributes such as the integrity and credibility of archive data, and it is
               difficult to reveal the characteristics of archive resources from different angles. Therefore, “archive
               reorganizing” from DH perspective needs to adopt a multi-dimensional organization model that
               combines static association and dynamic aggregation.
                 (1) Static association. It is the main way to statically associate archive data by using knowledge
               organization models such as domain ontology to describe, organize and standardize archival data,
               forming associated data that can be exchanged, mapped and inter-operated on various platforms
               of data of different structures under a specific framework. The archival data obtained from
               the “discovering” is still relatively isolated memory entities, which need to be associated and
               structured by the domain ontology. The people, events, times, places, physical entities contained
               in the archival data and related sources should be regarded as meaningful memory entities, and
               the semantic relationships between entities should be established in the form of triples through
               the object properties of the ontology. Since the categories and properties of the domain ontology
               are often constructed by domain experts, the structure and application method are relatively
               stable once determined, and it is not easy for the entire framework to change disruptively, as it
               follows the restrictions and specifications of the OWL language. Therefore, we call the association
               structure generated by the organization of archival data through the domain ontology “static
               association”, in order to express the transition of archival resources from “archival data” to “archival
               associated data”, which makes it possible to conduct the unstructured storage, open organization
               and associative publication of archival data.
                 (2) Dynamic aggregation. Compared with the static association of archival data relying on
               the ontology model construction, dynamic aggregation focuses more on revealing the essential
               attributes of archival resources from different dimensions, highlighting the difference between
               archival data research and other types of data research from DH perspective. The “dynamic
               aggregation” framework of archival data in this paper also requires the intervention of the
               ontology model, but the framework needs to be built top-down and bottom-up. First, it is necessary
               to build a top-down knowledge model, forming a multi-dimensional knowledge organization
               structure by establishing a “dimension model”. The “dimensional model” here mainly refers to
               the multi-dimensional classification system related to memory entities, which reorganizes the
               discrete, qualitative attributes that describe an entity. For example, the concept of “time” can be
               understood from different perspectives such as the era of the Common Era, feudal dynasties or
               other calendars with different attribute expressions while examples of feudal dynasties are some
               discrete, qualitative character values. Secondly, it is necessary to extract and fill the bottom-up
               instance extraction and filling process, that is, extracting the different attributes of the “discovering”
               memory entities in different dimensions and perspectives and filling them into the “dimension set”
               directly related to a concept. The conceptual instances of these dimensions can be preferentially
   159   160   161   162   163   164   165   166   167   168   169