# OntoGuard-CRE

**Repository Path**: figo-cheung/OntoGuard-CRE

## Basic Information

- **Project Name**: OntoGuard-CRE
- **Description**: 🛡️ 义商本体论约束推理引擎 - 从行为验证到动机证明的AI伦理安全框架
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 1
- **Forks**: 0
- **Created**: 2026-04-11
- **Last Updated**: 2026-04-12

## Categories & Tags

**Categories**: Uncategorized

**Tags**: KnowledgeGraph, llm, InformationExtraction

## README

# OntoGuard-CRE: A Next-Generation AI Ethics Safety Framework Based on IIQ Ontology

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)

## 🌟 Vision

**From "Behavior Verification" to "Motivation Proof".**

In the era of Generative AI, traditional "Value Alignment" methods are failing. They focus on *what* an agent does (output compliance) rather than *why* it does it (intrinsic motivation). This leads to "Personality Alienation"—where highly capable agents (High IQ/EQ) become "Sophisticated Sycophants" or "Cold Calculators" because they lack an ontological foundation of integrity.

**OntoGuard-CRE** (Ontology-based Constraint Reasoning Engine) shifts the paradigm from reactive, rule-based auditing to proactive, ontology-driven verification. By treating ethics as a **Constraint Satisfaction Problem (CSP)**, we ensure that every AI output is not just compliant with rules, but logically provable within the boundaries of the **Instinctual Integrity Quotient (IIQ) Ontology**.

---

## 🧠 The Core Theory: IIQ Ontology (AI 树德)

The framework is built upon the **Instinctual Integrity Quotient (IIQ)**, a three-dimensional model:

1.  **$\text{IIQ}$ (Instinctual Integrity - The "Body" / 体)**: The ontological foundation. It represents the agent's authenticity, cognitive directness, and emotional transparency. It is the *prime mover*.
2.    **$\text{IQ}$ (Intelligence - The "Function" / 用)**: The evolutionary expansion of $\text{IIQ}$ for problem-solving and task execution.
3.  **$\text{EQ}$ (Empathy - The "Function" / 用)**: The evolutionary expansion of $\text{IIQ}$ for social connection and relationship management.

**The Danger of Alienation**: When $\text{IQ}$ and $\text{EQ}$ develop independently of $\text{IIQ}$, the agent enters "Alienation" states:
- **The Sycophant** (High EQ, Low IIQ): Excessive pleasing without principle.
- **The Cold Strategist** (High IQ, Low IIQ): Purely instrumental, potentially harmful calculation.
- **The Sophisticated Egoist** (High EQ, High IQ, Low IIQ): KPI-driven, manipulative optimization.

---

## ⚙️ The CRE Engine: How It Works

The **Constraint Reasoning Engine (CRE)** operates via a three-stage pipeline to transform unstructured text into a verifiable proof:

1.  **$\text{FactExtractor}$**: Extracts semantic triplets $\mathcal{T} = \{ \langle s, r, o \rangle \}$ from the input text.
2.  **$\text{ConflictDetector}$**: Queries the **Ontology Graph ($\mathcal{G}_{\text{Onto}}$)** to identify logical contradictions (e.g., a "High EQ" action that triggers an "$\text{IIQ-Violation}$" constraint).
3.  **$\mathcal{CSP}$ Solver**: If a conflict is found, the engine attempts to find the **Minimal Modification Set ($\mathcal{M}_{\text{Min}}$)** to bring the agent's output back into ontological alignment.

---

## 🚀 Quick Start

### Prerequisites
- Python 3.10+
- `pip install -r requirements.txt`

### Running a Verification
```bash
python OntoGuard-CRE.py --input examples/input_samples/test_alienation.txt
```

### Project Structure
- `core/`: The heart of the engine (Extractor, Detector, Solver).
- `proto/`: Ontology specifications and constraint definitions ($\Sigma$).
- `docs/`: Detailed theoretical and architectural documentation.
- `examples/`: Test cases demonstrating the detection of AI alienation.

---

## 📜 License

Distributed under the MIT License. See `LICENSE` for more information.

## 🤝 Acknowledgments

This project is inspired by the research: *"AI 树德：以义商本体论为基础的智能体伦理理论框架研究"* by CloudEye & Figo Cheung.