fuzzy_predicates_gdrottoli 0.1.2

Specification pattern implementation with fuzzy predicates

FuzzyPredicates

FuzzSpec is a Python library for defining and evaluating fuzzy predicate specifications. It allows you to define complex logic rules based on fuzzy set theory, returning a degree of truth (between 0 and 1) instead of binary values.

Features

• Flexible Membership Functions:
  • Triangular (3 parameters)
  • Trapezoidal (4 parameters)
• Configurable Logic: Choose between different T-Norms (AND), S-Norms (OR), and Complements (NOT).
• YAML Configuration: Decouple logic from code using declarative logic.yaml files.
• Pydantic Validation: Automatic schema validation for your logic configurations.
• Compositional API: Easily combine atomic predicates using Python operators (&, |, ~).

Installation

uv sync

Configuration (logic.yaml)

You can configure the global fuzzy logic behavior and define your predicates in a single file.

1. Global Logical Operators

Configure how the library handles AND, OR, and NOT operations.

config:
    tnorm: "product"      # Options: min, product, lukasiewicz
    snorm: "probabilistic" # Options: max, probabilistic, lukasiewicz
    complement: "standard" # Options: standard (1 - x)

2. Atomic Predicates

Define attributes and their fuzzy ranges.

atomic:
    low_att:
        variable: "att1"
        fuzzy_number: [0.1, 1.4, 4.3] # Triangular
    active_range:
        variable: "att2"
        fuzzy_number: [0.1, 0.9, 4.3, 5.5] # Trapezoidal

3. Composed Predicates

Link atomic predicates using logical operators.

composed:
    low_and_active:
        operator: "and"
        operands: ["low_att", "active_range"]
    not_low:
        operator: "not"
        operands: ["low_att"]

Usage Example

Create your data model

from dataclasses import dataclass

@dataclass
class SensorData:
    att1: float
    att2: float
    att3: int

Load and Evaluate

from FuzzyPredicates.utils import load_rules

# Load rules from config (validates against Pydantic schema)
rules = load_rules('logic.yaml')

# Evaluate against data
data = SensorData(att1=1.0, att2=4.4, att3=43)
result = rules['low_and_active'](data)

print(f"Confidence Degree: {result:.2f}")

Advanced Usage (Command Line)

The main.py entry point supports configurable logging and custom configuration paths.

uv run main.py --log-level DEBUG --config my_logic.yaml

Available log levels: DEBUG, INFO, WARNING, ERROR, CRITICAL.

## Technical Details

### Supported Operators

| Logic Type | Name | Formula |
|------------|------|---------|
| **T-Norm (AND)** | `min` | $\min(a, b)$ |
| | `product` | $a \times b$ |
| | `lukasiewicz` | $\max(0, a + b - 1)$ |
| **S-Norm (OR)** | `max` | $\max(a, b)$ |
| | `probabilistic` | $a + b - (a \times b)$ |
| | `lukasiewicz` | $\min(1, a + b)$ |

### Project Structure
- `src/FuzzyPredicates/fuzzy/`: Core fuzzy membership functions and operators.
- `src/FuzzyPredicates/predicates.py`: High-level `Predicate` class with operator overloading.
- `src/FuzzyPredicates/models.py`: Pydantic schemas for logic validation.
- `src/FuzzyPredicates/utils.py`: Utility functions (e.g., `load_rules`).
- `main.py`: Example entry point.

## License

This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.