mondher 0.2.0

Formal Concept Analysis engine with Python bindings

mondher

Python bindings to the Mondher formal-concept-analysis engine.

Install

pip install mondher

Quickstart

import mondher

# Read a context file (Burmeister .cxt, CSV, DAT, XML, or JSON).
with open("breakfast.cxt", "rb") as f:
    ctx = mondher.parse_context(f.read(), filename="breakfast.cxt")

# Compute the concept lattice and get a list of dicts.
concepts = mondher.compute_concepts(ctx)
print(f"{len(concepts)} concepts")
for c in concepts[:5]:
    print(f"  extent={c['extent']}  intent={c['intent']}")

# Use a different algorithm. See `supported_algorithms()` for choices.
print(mondher.supported_algorithms())  # ['next-closure', 'in-close-5', 'fcbo']
concepts_fast = mondher.compute_concepts(ctx, algorithm="fcbo")

# Compute the Duquenne-Guigues canonical implication base.
implications = mondher.compute_implications(ctx)
for imp in implications[:3]:
    premise = ", ".join(imp["premise"])
    conclusion = ", ".join(imp["conclusion"])
    print(f"  {{{premise}}} -> {{{conclusion}}}")

Jupyter notebooks

In a notebook cell, just put a Lattice as the last expression and Jupyter renders it inline as SVG:

import mondher

ctx = mondher.parse_context(open("breakfast.cxt", "rb").read(), filename="breakfast.cxt")
lattice = mondher.compute_lattice(ctx)
lattice  # Renders the Hasse diagram inline

The Lattice class also has:

• .concepts() — returns the same list-of-dicts as compute_concepts
• .svg() — returns the SVG string (e.g., for saving to a file)
• .n_concepts — the number of concepts

Supported formats

The parse_context function dispatches on the filename extension:

Extension	Format
.cxt	Burmeister
.csv	Comma-separated values
.dat	NCI / dense binary
.xml	ConExp XML
.json	Mondher JSON

License

Proprietary. See the main repository for details.