Operationalizing Reliability Gaps in Large Language Models: A Semi-Systematic Evidence Map of Reasoning, Factuality, Evaluation, and Retrieval-Augmented Generation

This repository contains the supplementary documentation and machine-readable research data supporting a semi-systematic literature review and evidence map of large language model (LLM) reliability research. The review examines the inferential distance between observable capability signals and stronger reliability claims across four analytical areas: 1. reasoning faithfulness;2. factual grounding;3. evaluation validity; and4. evidence use in retrieval-augmented generation. The repository is intended to support reviewer inspection, study-level claim tracing, bibliographic verification, and methodological audit. It does not contain copyrighted full texts of the reviewed publications.