# Using own custom dataset in yente?

**URL:** <https://discuss.opensanctions.org/t/using-own-custom-dataset-in-yente/40>\
**Category:** Support & Questions\
**Tags:** support, yente\
**Created:** [February 19, 2025, 9:55am UTC](https://discuss.opensanctions.org/t/using-own-custom-dataset-in-yente/40 "2025-02-19T09:55:06Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![disco9283](https://avatars.discourse-cdn.com/v4/letter/d/c67d28/32.png) [@disco9283](https://discuss.opensanctions.org/u/disco9283)\
**Post date:** [February 19, 2025, 9:55am UTC](https://discuss.opensanctions.org/t/using-own-custom-dataset-in-yente/40/1 "2025-02-19T09:55:06Z")

</div>

Hey OpenSanctions team!

I’m looking for advice on how to load a non-OpenSanctions dataset into our on-prem yente instance.

We’ve got a CSV file that we want to update once a month that contains a block list of customers. Each row includes a person name, date of birth, tax ID and their customer ID.

How can I screen against this list alongside the main OpenSanctions data using yente?

---

<div class="post-metadata">

**Author:** ![pudo](https://dub1.discourse-cdn.com/flex014/user_avatar/discuss.opensanctions.org/pudo/32/1_2.png) [@pudo](https://discuss.opensanctions.org/u/pudo)\
**Post date:** [February 19, 2025, 12:53pm UTC](https://discuss.opensanctions.org/t/using-own-custom-dataset-in-yente/40/2 "2025-02-19T12:53:38Z")

</div>

Hi @disco9283, welcome here!

We’ve got some [existing documentation](https://www.opensanctions.org/docs/yente/datasets/) for this process, but perhaps a quick overview:

- You’ll need to **convert your data** to the FollowTheMoney data model [using a CSV or SQL mapping](https://followthemoney.tech/docs/mappings/). This is essentially a small YAML file that tells the FollowTheMoney command-line tool (`ftm`) how to convert each row in your data into a JSON object that `yente` will understand.
  - In order to do this, you need to install the `followthemoney` Python library somewhere so you can use it. If you’re good with creating a shell inside a Docker container (`docker run -ti --mount type=bind,src=/Users/pudo/Data,dst=/data ghcr.io/opensanctions/yente:latest bash`, you can use the copy that is pre-installed inside the yente container 🙂
  - Make sure your FtM dataset uses unique keys (entity IDs) that aren’t used by the main OpenSanctions dataset. By default, FtM generates keys as a hash of some of the columns of the source data, which will work well.

- Next, you’ll need to **make a manifest file for `yente`**. This is shown in the documentation, but the important thing here is that the manifest file needs to specify a path to your newly made FtM data.
  - The data path - alongside the manifest file - will need to be accessible inside your `yente` container, e.g. using a [bind mount](https://docs.docker.com/engine/storage/bind-mounts/) or as an `http` URL. If you’re using a bind mount, `yente` will use the modification time of the file to check if the data has been updated, so automatic updates should work as expected.
  - Inside the [manifest YAML](https://www.opensanctions.org/docs/yente/datasets/), you need to specify an extra dataset for your in-house data. You may also want to define an extra collection (ie. a dataset with a `children` section) that includes both that new dataset and the OpenSanctions `default` collection. This combined collection then becomes the scope for your API queries (eg. `/search/combinedscope`).

These are just some starting points - please feel free to share your mapping or any extra questions as you move forward!