How tartaruszen/anets works
Project type and purpose
• This is an open-source research tool (ANets) for efficiently summarizing attributed diffusion networks.
• It provides a command-line interface (CLI) in Python that orchestrates calls to a C++ core library to perform graph summarization and segmentation in parallel.
File structure
• Top-level:
– README.md (project description & usage)
– code/ (all source and build logic)
– data/ (sample “disney” and “memetracker” graphs)
• code/:
– ANets.py, getEigenScore.py – Python entry points (CLI wrappers)
– makefile – build script for C++
– src/ – C++ implementation modules:
• CoarseNet.* – summary-graph generator
• DisjointSets.* – union-find data structure for segmentation
• Edge.*, graph.* – core graph representations & I/O
• data/:
– Two example datasets, each with features.txt and links.txt
README for additional insights
• Workflow:
1. make → compile C++ into binaries/lib
2. make demo → run toy example end-to-end
3. python ANets.py <data_path> <percent> <threads1> <threads2> → real run
• Inputs: raw link list + node attributes
• Outputs: summarized links.txt, features.txt, node_map.txt
Detail
• Every major folder and module: code/src, code/ANets.py, data/<dataset>
• Output files (links.txt, features.txt, node_map.txt) as part of the Data Layer