v3.2.0

Note: this release includes a new version of scikit-learn, which required re-training the matching model; so the coefficients for the model are now different than before.

Mostly maintenance changes, although using the fingerprints will give better matches in particular for Russian company names.

What's Changed

• Bump nomenklatura from 2.7.5 to 2.7.7 by @dependabot in #193
• Bump fastapi from 0.88.0 to 0.90.1 by @dependabot in #210
• Bump pantomime from 0.5.3 to 0.6.0 by @dependabot in #205
• Bump docker/build-push-action from 3 to 4 by @dependabot in #203
• Bump orjson from 3.8.3 to 3.8.6 by @dependabot in #209
• Bump aiofiles from 22.1.0 to 23.1.0 by @dependabot in #207
• Bump email-validator from 1.3.0 to 1.3.1 by @dependabot in #200
• Bump types-aiofiles from 22.1.0.4 to 22.1.0.7 by @dependabot in #208
• Bump elasticsearch[async] from 8.5.3 to 8.6.1 by @dependabot in #201
• Bump aiohttp[speedups] from 3.8.3 to 3.8.4 by @dependabot in #212
• Bump fastapi from 0.90.1 to 0.91.0 by @dependabot in #211

Full Changelog: v3.1.3...v3.2.0

v3.1.2

This release includes numerous bug fixes related to the stability of the software, and fixes one backend issue with the matching API.

Full Changelog: https://github.com/opensanctions/yente/compare/v3.0.0..v3.1.2

v3.0.0

This release revamps the way we handle dataset metadata in yente in order to make it easier to load and update non-OpenSanctions datasets. It is a breaking change because it modifies the format of the manifest file (manifest.yml). If you're running yente with the default manifest file, the changes will not affect you.

Updating your manifest.yml

In order to update your custom manifest for yente 3.0.0, you need to change the external section into a catalogs section like this:

schedule: null
# old:
external:
  url: "https://data.opensanctions.org/datasets/latest/index.json"
  scope: all
# new:
catalogs:
  - url: "https://data.opensanctions.org/datasets/latest/index.json"
    scope: all
    resource_name: entities.ftm.json
datasets: []

The cool thing about this is that you can now add extra catalogs. Try putting in https://data.opensanctions.org/contrib/graph/catalog.json and it'll index a bunch of non-OpenSanctions data :)

What's Changed

• Revamp metadata architecture by @pudo in #188
• Bump structlog from 22.1.0 to 22.3.0 by @dependabot in #184
• Bump orjson from 3.8.2 to 3.8.3 by @dependabot in #189
• Bump nomenklatura from 2.7.4 to 2.7.5 by @dependabot in #191
• Bump elasticsearch[async] from 8.5.2 to 8.5.3 by @dependabot in #192

Full Changelog: v2.2.1...v3.0.0

v2.2.0

This release includes an updated/improved entity matching model and will produce different scores than the previous release.

What's Changed

• Bump fastapi from 0.85.0 to 0.85.1 by @dependabot in #158
• Bump orjson from 3.8.0 to 3.8.1 by @dependabot in #160
• Bump aiocsv from 1.2.2 to 1.2.3 by @dependabot in #161
• Bump followthemoney from 3.0.8 to 3.0.10 by @dependabot in #166
• Bump elasticsearch[async] from 8.4.3 to 8.5.0 by @dependabot in #167

Full Changelog: v2.1.0...v2.2.0

yente 2.1.0

This release changes the way that redirects for merged entities are performed. This should lead to a smaller index size and faster indexing.

yente 2.0

Yente 2.0 removes the statement API (used only to power the web site) and simplifies the overall codebase. It also significantly improves the documentation of the request endpoints in the OpenAPI docs.

1.8.0 - Stability improvements

This release includes the following:

• Stability improvements to the API and indexer, improving the performance and stability especially for large indexing jobs (30GB+).
• Updated scoring model for the matching API, including new model features and better precision.
• Updated data model (FollowTheMoney 3.0), with Address entities becoming first-class search targets.
• Various dependency updates and performance improvements.