This project demonstrates the application of GPU-accelerated graph analytics through RAPIDS cuGraph on a Facebook dataset provided by the Stanford Network Analysis Project (SNAP). By leveraging cuGraph functionalities, we perform various graph analytics methodologies including calculating degrees, PageRank, Betweenness Centrality, and community detection using the Louvain method.
The dataset represents a social network from Facebook. It consists of 'circles' (or 'friends lists') from Facebook. Facebook data was collected from survey participants using a Facebook app. The dataset includes node features (profiles), circles, and ego networks.
Before you begin, ensure you have met the following requirements:
- NVIDIA GPU
- CUDA and cuDNN installed and configured
- RAPIDS suite installed (cuDF and cuGraph libraries)
- Python 3.x
- NetworkX and Matplotlib Python libraries installed
- The Facebook dataset from SNAP
The script performs the following operations:
- Data Loading: Loads the Facebook dataset and initializes a cuGraph graph.
- Degree Calculation: Calculates the degree of each node in the network.
- PageRank Calculation: Computes the PageRank scores for each node, highlighting influential users in the social network.
- Betweenness Centrality Calculation: Computes the betweenness centrality scores for each node, identifying bridges or connectors in the social network.
- Community Detection: Applies the Louvain method for community detection, categorizing users into distinct communities, and calculates the modularity score.
- Visualization: Renders the network graph, color-coded based on community structure, using NetworkX and Matplotlib. It showcases the communities and highlights the relationships between users. The second and third graphs do the same but resize the nodes based on page rank and betweenness centrality, respectively.
