Knowledge Graph Embedding in E-Government

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The increasing adoption of Knowledge Graphs (KGs) in e-Government presents new opportunities for improving knowledge representation, inference, and decision-making in digital governance. However, real-world knowledge graphs often suffer from incompleteness, limiting their effectiveness in tasks such as policy analysis, citizen engagement, and service integration. To address this challenge, we explore Knowledge Graph Embedding (KGE), a Machine Learning (ML) technique, to generate meaningful vector representations of entities and relationships within an RDF/OWL knowledge graph developed using Protégé. Specifically, we apply TransE, a simple yet effective translational embedding model, to encode structured e-Government data and predict missing links. Our approach involves preprocessing the knowledge graph using RDFLib, extracting entity-relation triples, and leveraging PyKEEN for embedding computation and evaluation. The dataset is divided into 80% training, 10% validation, and 10% test sets, and model performance is assessed using standard metrics, including Mean Rank (MR), Mean Reciprocal Rank (MRR), and Hits@k (k = {1, 3, 10}). Experimental results reveal structural patterns within the e-Government knowledge graph, highlighting key entity clusters, relationship distributions, and link prediction trends. Furthermore, our analysis indicates that tail entity prediction tends to be more reliable than head entity prediction, suggesting areas for further optimization. Although TransE demonstrates promising results, challenges such as high variance in ranking metrics, dataset sparsity, and potential biases in entity distribution remain. Future research directions include hyperparameter tuning, integration of external ontologies, and the incorporation of neurosymbolic AI techniques to enhance semantic reasoning. This study contributes to the advancement of knowledge representation in e-Government, showcasing the potential of KGE to support intelligent decision-making, improve service delivery, and foster a more data-driven public administration. Keywords: Know