Introduction

The Song (宋) dynasty (960–1279) in Chinese ancient history has developed one of the most complex bureaucratic systems in premodern history. Its complexity stemmed not only from the separation of officials’ formal titles, actual duties, and salaries, but also from its continuous adjustments to the entire huge official system (see Figure 1). This complexity poses challenges for historical research and digital analysis.

Figure 1 The Song Dynasty Official System after the Yuanfeng Reform (1082)

Extant research has provided substantial foundations for the study of Song dynasty official system (abbreviated as the Song system). Dictionary of Song Dynasty Official System (Gong, 2017) offers the most comprehensive catalog of officials and agencies in the Song system. Several studies have examined institutional evolution and functional logic of the Song system (Deng, 1993; Zhao, 2010; Miao, 2022). From a digital humanities perspective, the China Biographical Database (CBDB) (Harvard University et al., 2025) provides metadata of Song dynasty officials, yet their representation of the Song system remains incomplete, especially regarding temporal attributes and functional transformations of official positions, as well as institutional evolutions. Hence, there is an absence of a structured and comprehensive representation of the Song system.

Our research designs a temporal knowledge graph for the Song dynasty official system, encompassing both official positions and agencies. We construct an ontology to make this complex bureaucratic system interpretable, searchable, and visualizable, showcasing how digital technology can aid the study of China’s institutional history.

Method and Process

We integrate standard frameworks including CIDOC CRM, GeoNames, and FOAF with information of the Song system, developing a domain-specific hierarchy (see Figure 2/3). To capture the dynamic nature of institutional evolution, we design delicate object properties (see Figure 4) including institutional creation, abolition, merging, restructuring, and functional reassignment. A particular innovation of our model is the distinction between  Official Time-span (the existence period of a position or institution) and Official Function Time-span (the duration for which a duty is attached to that position) to capture the evolution of official positions. Furthermore, we introduce specific properties such as Restructured into and Reset to to describe the evolution of institutions in Song dynasty.

After designing the ontology structure, we utilize Protégé 5.5.0 ontology builder to construct the Song system ontology model and generate a structured semantic network (see Figure 5).

Figure 2 Table of Hierarchical Relationships Between Ontology Categories of Official Positions in the Song Dynasty

Figure 3 Table of Hierarchical Relationships Between Categories of Song Dynasty Institution Ontology

Figure 4 Object Attribute Relationships

Figure 5 The Ontology of the Song Dynasty Official System

For data collection, we first identify the essential entity types and attributes required to represent Song official position, including name, function, and temporal attributes. On top of that, we extract data primarily from Dictionary of Song Dynasty Official System (Gong, 2017). All data entries are manually entered and normalized to ensure standard consistency. We collect 3,667 government official positions of Song dynasty. The resulting dataset provides a structured foundation that integrates attribute information, functional descriptions, and temporal data, which are critical for modeling the dynamic evolution of Song system.

We adopt Neo4j to develop the Song dynasty official system knowledge graph, following a workflow including node creation, attribute configuration, relationship establishment, and data import. First, information is extracted from organized data files to create independent nodes for each official position with corresponding attributes. Nodes for other entities (e.g., official duties, affiliated institutions) are also established with appropriate default values. In the relationship establishment phase, based on the relational attributes defined during the ontology construction, multiple relationships (e.g., possesses [power], undertakes [duties], belongs to [institutions], serves at [locations]) are built between official position nodes and other key entities. This clearly illustrates the associations between official positions and related entities. Finally, Neo4j’s Cypher query language is used to load processed data into the graph database. Nodes and relationships are created via CREATE statements, while potential duplicate data is handled using MERGE statements. This completes the knowledge graph construction, enabling efficient storage and intuitive visualization of official position information.

Leveraging Neo4j’s query capabilities, multi-dimensional queries are supported—Users may retrieve results through exact or fuzzy searches, trace institutional evolution across time, and explore the interconnectedness of positions, duties, and agencies.

Figure 6The Song Dynasty Official System Knowledge Graph

Conclusion

Our research demonstrates how digital humanities methods can be used to model the semantic and temporal complexity of premodern official system. The structured representation of 3,667 entities enables scholars to examine official functions, institutional transformations, and temporal attributes with precision and detail that traditional methods cannot easily provide. As for institutional information, our knowledge graph integrates evolutional institutional relationships and institution-position relationships, thereby analyzing the Song system from various angles, and provide users with both distant reading and close reading angles. We believe this knowledge graph will help historians broaden their analytical perspectives and open new avenues for the study of institutional history.

We have built a dedicated website for the project. However, due to the high cost of server maintenance, we are unable to keep the website running continuously. In the future, we will keep seeking appropriate opportunities to relaunch the website and continuously improve the dataset. It is our hope that our research findings can be extended to other digital humanities studies on the official system of ancient China.