matchlab 0.1.3

A local-first library for building, running and evaluating entity resolution pipelines.

Record matching is a chore. matchlab makes it a pipeline you can build, run, query and measure — on your machine, against your warehouse, with nothing to deploy.

import matchlab as mb

companies = mb.read_database(
    name="crn",
    sql="select pk, company, town from companies",
    client=warehouse,
    key_field="pk",
)

entities = (
    companies.clean({"name": "lower(crn_company)"})
    .dedupe(
        model_class=mb.NaiveDeduper,
        model_settings={"unique_fields": ["name"]},
    )
    .resolve()
    .collect()
)

entities.lookup_key(from_source="crn", to_sources=["dh"], key="a1")

Read the full documentation.

What it does

• A lazy plan. Source(...).dedupe(...).resolve() builds a tree of steps. Nothing runs until you collect().
• Content-addressed caching. Re-collecting an unchanged plan does no work. Adding a step runs only that step.
• Materialised resolver output. A collected resolver writes a complete (root, leaf, key, source) table, so lookups are reads, not re-derivations.
• Measurement as a first-class job. Sample clusters, record judgements, score precision and recall, and compare methodologies on equal terms.

What it doesn't do

No server, no accounts, no permissions, nothing to deploy. If you need a shared, governed matching service, matchlab is not that.

Installation

pip install matchlab

Coming from Matchbox?

matchlab is the successor to matchbox-db, with the server removed and the client API rebuilt. It's a hard break — see the migration guide.

Development

See our full development guide and coding standards on our contribution guide.