Entropy-Guided Adaptive Label Propagation for Location-Aware Graph Clustering

This repository provides a complete implementation of Entropy-Guided Adaptive Label Propagation for Location-Aware Graph Clustering.

• Adaptive Label Propagation (ALP): A label propagation algorithm that uses entropy-guided weighting to adjust the weights of structural and location information.
• Fixed Label Propagation (FLP): A label propagation algorithm that uses a fixed α value to combine structural and location information.
• Label Propagation (LP): A classic label propagation algorithm that uses majority voting.

The proposed ALP algorithm shows better performance than both FLP and LP algorithms in location-aware graph clustering tasks.

Code Structure

├── code/
│   ├── model/
│   │   ├── lp.py                   # Label Propagation (Majority Vote)
│   │   ├── flp.py                  # Fixed Label Propagation (Fixed α)
│   │   ├── alp.py                  # Adaptive Label Propagation (Adaptive α)
│   │   └── label_propagation.py    # Label Propagation implementation from NetworkX
│   ├── dataset.py                  # Dataset loading
│   ├── utils.py                    # Utility functions
│   └── main.py                     # Entry point
├── datasets/                       # Datasets (Brightkite, Gowalla, Custom)
│   │   ├── custom/            
│   │   │   ├── custom_edges.txt                
│   │   │   ├── custom_totalCheckins.txt
│   │   │  # Download from the link at the bottom
│   │   ├── Brightkite_edges.txt            
│   │   ├── Brightkite_totalCheckins.txt               
│   │   ├── Gowalla_edges.txt          
│   │   └── Gowalla_totalCheckins.txt  
└── README.md

How to Run

Requirements

• Python 3.8+

pip install -r requirements.txt

Example Usage

python main.py [-h] [--algorithm ALGORITHM] [--alpha ALPHA] [--dataset DATASET] [--output OUTPUT]

python main.py --algorithm LP --dataset brightkite
python main.py --algorithm FLP --dataset brightkite --alpha 0.5
python main.py --algorithm ALP --dataset brightkite

Arguments

Parameter	Description
--algorithm	Algorithm to use (LP, FLP, ALP)
--alpha	α value to use in FLP algorithm (0~1)
--dataset	Dataset to use (brightkite, gowalla, custom)
--output	Directory path where result files will be saved (default: ../result)

Datasets

• Brightkite
• Gowalla
• Custom