Asset- and Time-Aware Evidence-to-Control Mapping and Contradiction Detection for OT Cybersecurity Assessment

This paper proposes an asset- and time-aware methodology for evidence-to-control mapping and contradiction detection in Operational Technology (OT) cybersecurity assessment. The proposed methodology focuses on structured evidence-to-objective mapping, asset and temporal context analysis, contradiction classification, and provenance-constrained assessment generation. The study proposes a comparative evaluation of three approaches: non-structured Large Language Model (LLM) assessment, conventional Retrieval-Augmented Generation (RAG), and a structured, provenance-aware framework. The research uses NIST SP 800-82 Revision 3 as its primary OT security reference. A benchmark design of 60 base scenarios and five evidence conditions per scenario is proposed. Research status: This is a methodology-focused proposed-framework preprint. Dataset construction, independent annotation, implementation validation, and quantitative experimental evaluation remain pending. No empirical performance improvements or statistical findings are claimed.