How to select sites containing Sample documents in Purview Trainable classifiers

Question

Hello.
I am working on creating custom trainable classifiers, but I am unable to add new sites or folders containing my sample documents. I am only able to select from the list of sites and folders in the drop down and the site I want is not listed there.

Also, my goal is to use trainable classifiers to train my documents that are to be kept permanently so i will be able to detect them anywhere within the tenant and apply a permanent retention label on them. my question here is, is trainable classifier the best tool for this identification?
I look forward to all your responses and help.
Thanks.

Answer 1

Hi Gift Nwokoloh

It sounds like you're having trouble adding new sites or folders for your sample documents while creating custom trainable classifiers in Microsoft Purview. Let's break down your queries.

Adding New Sites or Folders

From your description, if your desired site isn't appearing in the dropdown list, it usually means that the site may not be indexed yet or there might be a permissions issue. Here are a few steps you can take:

Check Site Accessibility: Ensure that the account you are using has appropriate permissions to access the site and folders you want to use. Sometimes, if you’re not a member or don’t have access rights, the site may not show up.

Indexing Time: If you just created the site or folder, it typically takes some time for it to be indexed. Ideally, allow about an hour for new sites to appear in your options.

Folder Structure: Make sure that the seed content is organized properly into dedicated folders and that each folder holds only the relevant sample documents. Each folder should also be limited to either positive or negative seed content.

Trainable Classifiers for Permanent Retention

Regarding your question on whether trainable classifiers are the right tool to identify documents for permanent retention, the answer is generally yes. Trainable classifiers can effectively identify specific types of content, which you can utilize to apply retention labels. Here’s how it works:

• Once your classifier is trained with enough representative samples and you've tested it, you can use the outcome to classify documents across your tenant. This is ideal for identifying and applying permanent retention policies.

Additional Steps:

Training Process: Make sure you collect a suitable number of seed content items (50 to 500 for positives, and 150 to 1,500 for negatives) and wait for the status of your classifier to change from "In Progress" to "Training is complete" before using it.

Consult Documentation: Check out the official documentation for in-depth guidance on setting up and using trainable classifiers.

Follow-Up Questions

To better assist you, it would be helpful to know:

1. Have you confirmed that you have the correct permissions to access the site you want to add?
2. When did you create the new site or folder, and have you waited for a reasonable indexing period?
3. Can you clarify the types of documents you're looking to classify for permanent retention?

I hope this helps you get on track! If you need further assistance, feel free to ask.

References:

• Get started with trainable classifiers
• Classification best practices in the Microsoft Purview governance portal
• How to apply classifications
• Manually apply classifications on assets in Data Map
• Resolve issues with classification and sensitivity labels - missing or incorrectly classified assets 

Answer 2

Welcome to Microsoft Q&A

Hello Gift Nwokoloh,

I can help with both the technical issue and the strategy question.

You can't see your site is usually because the site is too new.

The list you see in Purview isn't live; it relies on a background search index. If you just created the folder or site today, Purview hasn't "found" it yet.

Wait at least 1 to 24 hours. It will eventually show up. Also, double-check that your user account is an Owner of that SharePoint site.

Honestly, using a Trainable Classifier for Permanent Retention is risky.

Trainable Classifiers are like teaching a human—they make mistakes. They might miss a file (meaning it could get deleted later) or tag the wrong file (meaning you keep junk forever).

The "Bucket" Method: create a specific Folder or Document Library for these important files. Go to "Library Settings" and set a Default Retention Label on that folder.

• Any file anyone drops into that folder gets the "Permanent" label instantly. It is 100% accurate and you don't need to train a complicated AI model.

Use Trainable Classifiers only if you have absolutely no way to organize the files into specific folders.

😊 If my answer helped you resolve your issue, please consider marking it as the correct answer. This helps others in the community find solutions more easily. Thanks!

Take a look at this video Get Started with Trainable Classifiers in Microsoft Purview in 30 minutes! This video is relevant because it walks through the setup of Trainable Classifiers and discusses the "seeding" process where you are currently stuck.