Best for
Frequent Subgraph Mining Workflows
Use `submine` when you want one consistent interface over multiple graph mining backends.
Tool
submine
A research-grade library for frequent subgraph mining with a unified interface over multiple mining backends and graph formats.
Stack
Python + Mining Backends
Built for Python-first workflows while handling the backend-specific friction for you.
What It Does
submine helps users focus on what to mine instead of manually adapting graph data for each
mining backend. It validates inputs, detects supported graph formats, and routes execution
through a unified Python API.
- Unified interface across multiple mining algorithms
- Automatic handling of supported graph formats
- Safer and more reproducible execution flow
- Designed for research and extension
Supported Inputs
- Edge lists such as `.txt` and `.edgelist`
- gSpan datasets such as `.data` variants
- Single-graph `.lg` files
- GEXF graph files
Installation
pip install submineFor development work:
pip install -e ".[dev]"Example Usage
from submine.api import mine_subgraphs
results = mine_subgraphs(
data="graphs.data",
algorithm="gspan",
min_support=5
)Typical Use Cases
- Discover repeated substructures across graph datasets
- Analyze patterns in social, biological, or information networks
- Prototype graph mining workflows without rewriting formats by hand
Why It Helps
- Reduces repetitive data conversion work
- Keeps graph-mining experiments easier to compare and reproduce
- Makes it simpler to move from small tests to broader mining studies
Further Reading
Why Subgraph Mining Matters
I wrote a companion article on why subgraph mining matters and how it shifts attention from isolated nodes to recurring structural patterns in complex networks.