Bibliometric Analysis and Co-word Mapping: The Field of Knowledge Graphs

Purpose: The rapid expansion of research in the field of Knowledge Graphs (KGs) over the past decade has positioned The field as a dynamic area at the intersection of computer science, artificial intelligence. Following Google’s introduction of the KG in 2012, scholars and industry stakeholders have increasingly explored the conceptualization, development, and application of KGs across a wide range of disciplines. KGs, as structured representations of entities, relationships, and semantic information, enable machines to process heterogeneous data in a contextually meaningful and interpretable manner. Beyond traditional knowledge representation, new directions such as ontology modeling, semantic web integration, and graph neural networks (GNNs) are driving innovation. Despite these advances, large-scale, systematic, and global evaluations of the intellectual structure, collaboration patterns, and thematic clusters in the field of KGs remain limited. This study addresses this gap by examining the evolution of research in the field of KGs between 2013 and 2025 using bibliometric and scientometric approaches. The overarching purpose is to uncover research trends, identify leading contributors and institutions, explore co-authorship and co-word networks, and highlight conceptual clusters shaping the KG landscape.
Methodology: This applied-descriptive bibliometric research was conducted using the co-word analysis method .Data were collected from the Scopus citation database, covering outputs published between January 2013 and July 2025 with the keyword "knowledge graph" in titles, abstracts, and author keywords. Keyword normalization reduced 19141 initial keywords to 19058 unique terms. After removing duplicates, excluding irrelevant records, and normalizing author, institution, and country labels, a curated dataset of 9000 articles was obtained. Furthermore, to enhance the accuracy of the analyses, keywords and knowledge graph-related concepts were carefully categorized and standardized. Data analysis and visualization were conducted using VOSviewer, mapping collaboration networks and co-occurrence clusters. The study examines temporal publication and citation trends, prolific countries, funding bodies, influential authors, highly cited works, and thematic clusters, with special attention to knowledge graph embedding, recommender systems, semantic web applications, and natural language processing.
Findings: Results showed that research in the field of KGs has grown substantially, reaching a publication peak in 2024 with 2541 articles and 33701 citations. Asia contributes the majority of publications (83.7%), with China accounting for 6,889 articles (76.5% of the total). The United States ranks second with 987 publications, followed by the United Kingdom, Germany, and Australia. This reflects strong investment by Chinese funding bodies—particularly the National Natural Science Foundation of China (NSFC) and the National Key Research and Development Program. Leading publication outlets include IEEE Access (345 articles), Knowledge-Based Systems (293 articles), and Applied Sciences Switzerland (248 articles), with many top journals in the Q1 quartile of Scopus. At the author level, Markus Kraft (University of Cambridge) leads with 48 publications, while Chinese researchers dominate the top ten. Co-authorship analysis revealed seven author clusters (42 nodes, 62 links) and wide participation from 136 countries, with China, the US, and Germany serving as central hubs of international collaboration. Co-word analysis identified six major clusters, including graph neural networks, deep learning, ontology, KG embeddings, question answering, and large language models (LLMs). Highly cited works include comprehensive surveys on KGs published between 2017 and 2022 in leading computing journals, which have shaped the understanding of KG construction, refinement, and applications. . Emerging contributions, such as roadmaps on unifying LLMs and KGs, along with applications in digital twin systems and scientific entity recognition, illustrate the field’s ongoing shift toward integrating advanced AI paradigms.
Conclusion: The bibliometric analysis showed that research in the field of KGs has emerged as a rapidly growing interdisciplinary area, characterized by strong publication growth, high geographic concentration in Asia, and expanding international collaboration networks. Chinese institutions and scholars play a leading role, supported by significant policy and funding initiatives, while Europe and North America remain important contributors. Thematic clustering highlights both the consolidation of established domains (ontology, semantic web, knowledge representation) and the expansion into cutting-edge areas (graph neural networks, embeddings, digital twin, recommender systems, and LLM integration). The findings provide a comprehensive overview of the intellectual structure and evolving dynamics of research in the field of KGs, offering insights for future studies. Scholars can use this mapping to identify promising areas of inquiry, build international collaborations, and navigate the funding and publication landscape. Ultimately, KGs are shown to play a pivotal role in advancing intelligent systems, semantic technologies, and digital transformation across sectors such as healthcare, education, natural language processing, and recommender systems. The integration of bibliometric analysis with scientometric evaluation further emphasizes the maturity of this field and its potential for shaping next-generation knowledge-driven applications.