Enriching Taxonomies Using Large Language Models

This collection of recent arXiv papers centers on Retrieval-Augmented Generation (RAG), large/multimodal LLM capabilities for retrieval, and infrastructure/benchmarks enabling production and edge deployments. Key contributions include: Taxoria (LLM-based taxonomy enrichment with validation/provenance), RAGdb (single-file, ONNX/SQLite RAG stack for edge/air-gapped environments), several RAG/retriever improvements (SmartChunk, DS-Serve, adaptive prefiltering, neural retriever-reranker pipelines), domain-specific RAG evaluations (anatomical pathology protocols with a RAG assistant), large-scale/µ