User-Side Contextual Interaction Assessment (USCI): A Methodology Specification for Post-Interaction Contextual Risk Assessment in Human-AI Interaction

This deposit presents the User-Side Contextual Interaction Assessment (USCI) v1.0.0, a pre-empirical methodology specification for post-interaction, user-side contextual risk assessment in human-AI interaction.

USCI is the operational implementation layer of the User-Side Contextual Hallucination (USCH) theoretical framework and is built upon the CXC-7 and CXOD-7 conversational context models. The methodology defines a four-axis risk profile (Fact Reliability, Context Alignment, User-side Safety, System Usability), a three-level primary state classification, five risk subtypes, and dual-condition collapse flag logic, with mandatory evidence traceability requirements.

This deposit is a public-layer, pre-empirical methodology release. Reproducibility in this record refers to structural auditability (definitions, output schema logic, evidence-traceability rules, and responsible-use boundaries), not deterministic reconstruction of controlled internals. Controlled-access components are withheld to prevent adversarial gaming and are available to qualified researchers upon direct request to the author.

This deposit includes:
- USCI_Public_Methodology_v1.0.0_EN.pdf – Main methodology specification
- USCI_Methodology_v1.0.0_Official_Corrigendum_and_Compliance_Addendum_EN.pdf – Normative corrigendum and compliance addendum
- SHA256_CHECKSUMS.txt – File integrity verification