Can build content for custom classifier be specified using generic urls?

Abby Newcomb 0

I’m a bit confused by this APIdocumentation regarding how to train a custom document classifier in v4.0. The ClassifierDocumentTypeDetails specifies the build content files, and has 2 possible formats (azureBlobSource and azureBlobFileListSource) which both require the files to be stored in an Azure Blob container.

There are two other ContentSourceKinds listed, url and base64, but no information on how to specify the build content in that format.

Are those ContentSourceKinds, url and base64, available? If so, what is the spec for the format of the build content?

kayukholahola 0 Reputation points

2025-01-06T21:00:10.99+00:00

Operation-Location: https://myendpoint.cognitiveservices.azure.com/documentintelligence/operations/b704bb00-d130-4f3f-a1d8-ca96de3eabb4?api-version=2024-11-30
Abby Newcomb 0 Reputation points

2025-01-07T14:53:26.3333333+00:00

Thanks Vikram.

Does that mean that only one document per class can be provided in the training data if using urlSource or base64Source? How can I specify multiple documents for training a single class using urlSource or base64Source?
Vikram Singh 100 Reputation points Microsoft Employee

2025-01-08T04:28:56.2766667+00:00

To specify multiple documents for training a single class using urlSource or base64Source, you typically need to provide a list of URLs or base64-encoded strings in your request. While the documentation may not explicitly state how to handle multiple documents with urlSource or base64Source, you can typically structure your requests to include arrays of these sources.

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Vikram Singh 100 Reputation points Microsoft Employee

2025-01-09T13:01:44.3766667+00:00

Hello @Abby Newcomb , Greetings! Just following up to check if my suggestion helped. Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Vikram Singh 100 Reputation points Microsoft Employee

2025-01-13T06:31:10.02+00:00

Hi @Abby Newcomb ,

We haven’t heard from you on the last response and was just checking back to see if you got a chance to try above suggestions. If the response helped, please do click Accept Answer and Yes for was this answer helpful.

I highly appreciate your contribution to the community.

Thank You
Abby Newcomb 0 Reputation points

2025-01-13T14:51:49.3833333+00:00

I'm not seeing an accept answer button, but yes my question has been resolved, though I haven't tested this yet.

1 answer

Vikram Singh 100 Reputation points Microsoft Employee

2025-01-07T05:11:41.14+00:00

Hello @Abby Newcomb ,

Welcome to the Microsoft Q&A Forum! Thank you for your question.

Yes, the mentioned ContentSourceKinds available. To clarify, the ClassifierDocumentTypeDetails allows for two content source formats: urlSource and base64Source. You need to specify one of these formats for your document classification

• urlSource: This requires a URI pointing to the document.

• base64Source: This requires the document to be provided in Base64 encoding.

If you have any further questions, feel free to ask!
Please sign in to rate this answer.

1 person found this answer helpful.
Vikram Singh 100 Reputation points Microsoft Employee

2025-01-14T05:10:54.2766667+00:00

Hi @Abby Newcomb Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

Thank you
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

Can build content for custom classifier be specified using generic urls?

1 answer

Your answer