Digital transformation has become a central driver of public administration modernization, as governments increasingly rely on digital technologies to improve public service delivery, administrative efficiency, transparency, and citizen-oriented governance. However, the rapid expansion of digital government research has generated a fragmented literature, dispersed across concepts such as e-government, digital transformation, public services, governance, innovation, and emerging technologies. This fragmentation limits the understanding of how the field is thematically structured and how scientific collaboration is organized across countries and institutions. Therefore, this paper maps the intellectual structure of digital government literature and identifies the dominant themes, emerging research directions, and collaboration patterns within this field. A bibliometric science-mapping analysis was conducted on 1,877 documents indexed in the Web of Science Core Collection, retrieved using the topic search terms "digital transformation" and "public services". The analysis was performed with VOSviewer and focused on keyword co-occurrence, country coauthorship, and organizational co-authorship networks. The findings show that digital transformation serves as the conceptual hub of the field, connecting traditional e-government research with newer themes in innovation, governance, artificial intelligence, big data, trust, adoption, performance, and post-pandemic public service resilience. The countrylevel network reveals broad but uneven international collaboration, while the organizational network remains more fragmented. The study contributes by integrating thematic and collaboration-based perspectives, showing that digital government research is expanding, interdisciplinary, and globally visible, but still institutionally fragmented and unevenly connected. These findings indicate the need for stronger institutional partnerships, interdisciplinary integration, and more coherent research networks in future digital government studies

Path-dependent dynamic systems are often analysed through separate metric, stochastic, econometric or nonlinear methods, while their integration within a unified architecture remains limited. This study develops and validates a Hybrid Dynamic Modelling Framework (HDMF) for path-dependent stochastic systems, using EU-27 digital public service trajectories during 2017–2022 as an empirical application. The framework combines DTW-TimeSeriesKMeans clustering, Markov transition analysis, PVAR, SEM and MLP approximation. The findings answer the four research questions sequentially. First, EU digital public service development is represented by three distinct trajectory regimes rather than a single convergence path. Second, the Markov layer shows strong diagonal dominance, limited mobility and practical non-ergodicity over the finite observation horizon, indicating persistent state dependence. Third, the PVAR layer confirms dynamic stability and bounded feedback within the institutional–digital system. Fourth, the SEM and MLP layers identify governance as a coherent latent institutional structure and reveal nonlinear predictive heterogeneity, suggesting possible threshold-type behaviour. Overall, the findings support the HDMF as a coherent multilayer architecture for modelling complex path-dependent systems and demonstrate its empirical usefulness through the EU digital public service case.