ai-fuzzy-rec

AIFuzzyRec

A Python library for generating AI-powered item recommendations using fuzzy logic and SQLite databases. AIFuzzyRec leverages fuzzy set theory and artificial intelligence to compute item similarities based on user-item interactions, providing flexible and AI-enhanced recommendations for applications like e-commerce, content platforms, or personalized systems.

Features

• Fuzzy Logic-Based Recommendations: Uses fuzzy membership functions to compute item similarities, allowing nuanced recommendation scores.

• Flexible Database Schema: Supports any SQLite database with user-item interaction data, with customizable table and column names.

• Context Manager Support: Ensures automatic cleanup of database connections using Python's with statement.

• Configurable Fuzzy Weights: Adjust weights for low, medium, and high similarity to fine-tune recommendation behavior.

• Error Handling: Robust handling of invalid user/item IDs and empty databases.

• Logging: Optional logging of recommendations to a file for debugging and analysis.

• Easy Integration: Lightweight and compatible with Python 3.6+.

Installation

Install AIFuzzyRec using pip:

pip install ai-fuzzy-rec

Or, if using Poetry, add it to your project:

poetry add ai-fuzzy-rec

Requirements

Python >= 3.6 pandas >= 2.0.0 scikit-fuzzy >= 0.4.2 numpy >= 1.25.0 scipy >= 1.10.0 packaging >= 21.3

Usage

Basic Example

Generate recommendations using a SQLite database (purchases.db) containing user-item interactions:

from ai_fuzzy_rec import FuzzyRecommender

# Initialize with default settings
with FuzzyRecommender(db_path='purchases.db') as recommender:
    # Recommend items similar to item ID 2
    item_rec = recommender.recommend_items_for_item(2, top_n=3)
    print(f"Items recommended for item 2: {[(item, round(score, 3)) for item, score in item_rec]}")

    # Recommend items for user ID 1
    user_rec = recommender.recommend_items_for_user(1, top_n=3)
    print(f"Items recommended for user 1: {[(item, round(score, 3)) for item, score in user_rec]}")

Expected Output:

Items recommended for item 2: [(3, 0.5), (4, 0.5), (1, 0.5)]
Items recommended for user 1: [(4, 1.0), (5, 0.5)]

Custom Schema and Parameters

Use a custom database schema and adjust fuzzy weights:

from ai_fuzzy_rec import FuzzyRecommender

with FuzzyRecommender(
    db_path='custom.db',
    table_name='orders',
    user_col='customer_id',
    item_col='product_id',
    fuzzy_weights=(0.1, 0.4, 0.5),  # Emphasize high similarity
    log_file='custom_recommendations.log'
) as recommender:
    item_rec = recommender.recommend_items_for_item(2, top_n=2)
    print(f"Top 2 items for item 2: {item_rec}")

Database Setup

Create a SQLite database with user-item interactions:

import sqlite3

# Create a sample database
conn = sqlite3.connect('purchases.db')
cursor = conn.cursor()
cursor.execute('''CREATE TABLE IF NOT EXISTS purchases
                 (user_id INTEGER, item_id INTEGER)''')
# Insert sample data
data = [(1, 1), (1, 2), (1, 3), (2, 2), (2, 3), (2, 4),
        (3, 3), (3, 4), (3, 5), (4, 1), (4, 4), (5, 1), (5, 2)]
cursor.executemany('INSERT INTO purchases VALUES (?, ?)', data)
conn.commit()
conn.close()

Testing

Run tests to verify functionality:

poetry run pytest

This executes tests in tests/test_ai_fuzzy_rec.py, ensuring recommendations and error handling work as expected.

Project Structure

/home/m/pyprj/
├── ai_fuzzy_rec/
│   ├── __init__.py
│   ├── ai_fuzzy_rec.py
├── create_database.py
├── purchases.db
├── pyproject.toml
├── README.md
├── LICENSE
├── test_recommender.py
└── tests/
    └── test_ai_fuzzy_rec.py