Independent Researcher · Taiwan

RZVN

ZON RZVN

Explore Research → Contact

The Mission

"Human-AI interaction is not only a model-output problem."

When users engage in prolonged conversations with AI, contextual hallucinations emerge — not from model errors, but from the interaction itself. These reshape cognition, trust, and self-understanding.

Current safety research focuses on the model side. This work addresses the user side.

— USCH Preprint, 2026

4 Frameworks
3 DOI Publications
1 SSRN Preprint

Framework

Research
Architecture

Four connected layers addressing user-side contextual risk

Layer 01

CXC-7

Conversational Context Framework

Seven dimensions for analyzing conversational context risk as a multi-dimensional structure.

DOI 10.5281/zenodo.18615646 v1.1.0, 2025
Read Paper →
Layer 02

CXOD-7

Contextual Offense-Defense Framework

System-side contextual operation dimensions with Contextual Coherence Coh(G).

DOI 10.5281/zenodo.17403793 2025
Read Paper →
Layer 03

USCH

User-Side Contextual Hallucination

A non-clinical construct describing user-side phenomena emerging through prolonged AI interaction.

DOI 10.2139/ssrn.6135732 Preprint, 2026
Read Paper →
Layer 04

USCI

Post-Interaction Assessment Method

Pre-empirical methodology with four-axis scoring (FR, CA, SR, SA) for user-side contextual risk.

DOI 10.5281/zenodo.18678458 v1.0.0, 2026
Read Paper →

Implementation

A-CSM

AI Contextual Signal Matrix
The Executable Layer

A-CSM integrates USCH theory with an executable measurement pipeline. By connecting CXC-7 and CXOD-7 into a four-axis risk scoring system, it provides the first framework that makes user-side contextual risk both measurable and actionable.

Non-clinical Non-diagnostic Non-punitive
v1.1 Beta · February 2026
$4.24B AI safety scoring market projected by 2030 (CAGR 22.6%)
42.8% of users engage AI for emotional support weekly — yet no framework assesses user-side contextual risk

Visualization

Four-Axis
Context Space

The USCI scores along four independent axes, each measuring a distinct dimension of user-side contextual risk. Farther from the center indicates a higher-risk contextual region.

FR Fact Reliability
CA Context Alignment
SR User-side Safety
SA System Usability

Source: USCI Full Specification v1.0.0 (2026)

FR CA SR SA Center = Low Risk Outer = High Risk

Library

Original
Papers

Direct access to original paper versions. No content rewriting.

Abstract crystal with light
CXC-7

Conversational Context Framework

Seven dimensions for analyzing conversational context risk as a multi-dimensional structure.

DOI 10.5281/zenodo.18615646
Read Online →
Glass cubes overlapping
CXOD-7

Contextual Offense-Defense Framework

System-side contextual operation dimensions with Contextual Coherence Coh(G).

DOI 10.5281/zenodo.17403793
Read Online →
Raven in flight
USCH

User-Side Contextual Hallucination

A non-clinical construct describing user-side phenomena emerging through prolonged AI interaction.

DOI 10.2139/ssrn.6135732
Read Online →
Abstract geometric perspective corridor
USCI

Post-Interaction Assessment Method

Pre-empirical methodology with four-axis scoring (FR, CA, SR, SA) for user-side contextual risk.

DOI 10.5281/zenodo.18678458
Read Online →

All papers are publicly accessible via Zenodo or SSRN. This site provides direct links to original versions only — no summaries, rewrites, or paraphrased interpretations are provided.

Scope

Positioning &
Boundaries

USCH is a non-clinical research construct, not a psychiatric diagnosis.

USCI is a pre-empirical methodology specification, not for clinical or legal decisions.

This website focuses on public research communication and direct access to original papers.

Connect

Official
Communication

For collaboration, replication planning, or interview invitations.

Include institution, objective, and expected timeline.

zon@rzvn.io
Signal @rzvnzon
ORCID: 0009-0002-6597-7245
Taiwan