mutato

Ontology-driven synonym swapping for semantic text enrichment. Mutato identifies terms in input text and replaces them with semantically equivalent synonyms sourced from OWL ontologies, enabling consistent, structured analysis of natural language content.

Use Cases

• Normalize terminology across transcripts before downstream analysis
• Enrich tokens with ontology-backed synonym candidates
• Bridge informal language to structured vocabulary in NLP pipelines

Quick Start

from mutato.parser import owl_parse

results = owl_parse(tokens=["student", "learned", "math"], ontologies=[...])

Installation

make all

This downloads the spaCy model, installs dependencies, runs tests, builds the package, and freezes requirements.

Or step by step:

make get_model   # download en_core_web_sm
make install     # poetry lock + install
make test        # run pytest
make build       # install + test + poetry build
make freeze      # export requirements.txt

CLI

The parse command parses input text against an OWL ontology and prints canonical forms:

poetry run parse --ontology path/to/ontology.owl --input-text "fiscal policy analysis"

Three modes are available:

See docs/cli.md for the full reference, including the MIXED-schema caveat for --live.

Architecture

Mutato is organized into four modules:

See docs/architecture.md for design details.

Matching Strategies

The parser applies multiple matching passes in order:

1. Exact -- literal string match against ontology terms
2. Span -- multi-token window matching
3. Hierarchy -- parent/child concept traversal
4. spaCy -- lemma and POS-aware NLP matching

Requirements

• Python >= 3.10, < 3.14
• Poetry for dependency management
• spaCy en_core_web_sm model (installed via make get_model)

Links

• Issue Tracker
• Source