Title
=====
Semantic Enhanced Clothing Recommendation System (ATSRS)

Description
===========
This project implements the Ant Trust-based Semantic-enhanced Recommendation System (ATSRS), 
a dynamic recommendation framework that integrates semantic item representation, trust networks, 
and Ant Colony Optimization (ACO) to improve personalized clothing recommendations.

Dataset Information
===================
- **Dataset**: Amazon Fashion (men's/women's shoes, bottoms, shirts)
- **Source**: https://registry.opendata.aws/amazon-reviews/
- **Size**: 45,184 users, 166,270 items, 358,003 entries
- **Sparsity**: 99.9952%
- Preprocessing:
  - Removed users with <5 interactions
  - Normalized ratings (min-max scaling)
  - Random split: 75% train, 25% test

Code Information
================
Structure:
src/
  data/               # Loaders, preprocessing
  models/             # Recommendation models (semantic, trust, ACO)
  utils/              # Metrics, configs, logging
scripts/              # Data prep, evaluation scripts
configs/              # YAML config files

Usage Instructions
==================
1. Install requirements:
   pip install -r requirements.txt
2. Prepare dataset:
   python scripts/prepare_data.py --raw_dir data/raw --out_dir data/processed
3. Run main pipeline:
   python src/main.py --config configs/default.yaml
4. Evaluate results:
   python scripts/eval.py --preds runs/<run-id>/preds.csv --truth data/processed/test.csv

Requirements
============
- Python >= 3.8
- numpy, pandas, scikit-learn
- networkx, owlready2
- pyyaml, tqdm, matplotlib

Methodology
===========
Offline Phase:
- Semantic item representation using OWL ontology
- K-Medoids clustering for semantic grouping
- Implicit Trust Graph construction from user-item interactions
Online Phase:
- Contextual trust-based neighbor selection using ACO
- Top-N recommendations ranked by semantic similarity & trust
- Dynamic pheromone update based on user feedback

Citations
=========
If you use this work, please cite:
Luyao He, Huazhi Xiang, Pietro Alex Marra, Chunjia Wang. 
"A Semantic Enhanced Clothing Design System based on implicit trust graphs and Ontology web language", 2025.