An Uncertainty-Gated Neuro-Symbolic Framework for High-Coverage Topic Modeling and Trend Analysis in Scholarly Corpora With LLM Assistance

The rapid growth of scientific literature demands scalable methods that can track research evolution, yet density-based topic models such as BERTopic systematically exclude low-density documents as outliers, obscuring emerging and niche research areas.We propose aNeuro-Symbolic, Uncertainty-Gated Frameworkthat recovers these outliers through geometric centroid reassignment and an ontological entropy gate derived from the Computer Science Ontology (CSO), routing only genuinely ambiguous cases to a local Large Language Model (Qwen2.5-14B via Ollama, temperature=0.0). A controlled ablation study demonstrates that centroid reassignment provides the largest coverage gain (+22.9 percentage points (pp)), the CSO entropy gate preserves niche-topic integrity, and selective LLM routing adds an additional +5.9 pp. On 12,535 Turkish computer engineering theses (2001–2025), the full pipeline raises coverage from 74.6% to 95.4% while maintaining competitive coherence (NPMI=0.100) and stability (AMI=0.832), at ∼15×fewer LLM calls than a fully generative Pure-LLM baseline. Mann–Kendall trend tests on the high-coverage series identify 69 statistically significant trends (FDRq< 0.05), and cross-corpus validation on 200K arXiv CS abstracts confirms that the architecture generalizes beyond the primary dataset. The framework offers a reproducible, cost-effective solution for monitoring scientific developments in rapidly evolving fields.