KNOWLEDGE GRAPHS IN DIGITAL HUMANITIES: A SEMANTIC FRAMEWORK FOR CULTURAL KNOWLEDGE DISCOVERY

Abstract

Purpose: This study proposes a semantic knowledge graph framework for cultural knowledge discovery in digital humanities using the China Biographical Database (CBDB). The framework supports semantic organization, relationship interpretation, and evidence-based knowledge discovery from large-scale historical records.

Methodology/Approach: A design science and secondary-data methodology was adopted. The framework integrates relational profiling, semantic modeling, spatial concentration analysis, and multiplex network analysis to construct and evaluate a semantic knowledge graph. Nine summaries of the complete database (659,593 persons) and five analytical subsets (5,000 persons) were examined using degree, strength, PageRank, approximate betweenness, Louvain community detection, Spearman correlations, and sensitivity analysis.

Findings: The semantic knowledge graph demonstrates broad relational coverage while revealing temporal incompleteness and uneven information distribution. The analytical graph contains 18,810 semantic relationships and 359 knowledge communities. PageRank shows moderate consistency with heterogeneous record prominence (ρ = .465), with stable rankings across weighted specifications but noticeable variation within the kinship layer. These findings demonstrate the effectiveness of semantic network analysis for uncovering implicit cultural knowledge and structural patterns in historical datasets.

Originality/Relevance: The proposed framework extends knowledge graph applications in digital humanities from visualization to semantic knowledge discovery. By emphasizing semantic organization, relationship analysis, and methodological transparency, it provides a reproducible approach for cultural analytics, historical network research, and digital humanities scholarship.