<![CDATA[The rapid proliferation of Learning Management Systems (LMS) in K–12 education has generated a substantial body of research, yet how its core themes emerge, converge, and transform over time remains insufficiently understood. Existing bibliometric and topic modeling approaches produce static snapshots of the literature, structurally incapable of capturing the dynamic epistemic processes through which research communities form and evolve. This study introduces Dynamic Semantic Network Analysis with Evolutionary Community Detection (DSNA-ECD), a novel computational framework that conceptualizes the K–12 LMS research field as a living epistemic system — a conceptual reframing that constitutes a distinct contribution to the K–12 LMS literature beyond prior static approaches. DSNA-ECD integrates three methodologically principled components: transformer-based semantic embeddings via Sentence-BERT (`all-MiniLM-L6-v2`), selected for its capacity to capture latent semantic proximity beyond lexical co-occurrence; a hybrid weighting scheme empirically calibrated to balance structural and semantic network signals; and the Leiden algorithm for community detection, preferred over Louvain for its theoretical guarantee of well-connected partitions and superior modularity optimization. Applied to a two-decade corpus of K–12 LMS publications, findings reveal a maturing field progressing from exploratory fragmentation through consolidation toward sophisticated integration of AI-enhanced adaptive systems and learning analytics. Compared to co-citation analysis, LDA topic modeling, and static semantic networks, DSNA-ECD uniquely offers semantic depth, guaranteed community coherence, calibrated hybrid weighting, and full cross-temporal trajectory tracking. Critically, findings reveal urgent underrepresentation of equity, algorithmic transparency, and ethical deployment research as AI-enhanced LMS systems proliferate, with direct implications for researchers, educational technologists, and policymakers.]]>
Copyrights © 2026
