Share via

How to rebuild Azure Content Understanding analyzer

Pavlo 0 Reputation points
2026-05-22T09:03:44.4433333+00:00

Article https://learn.microsoft.com/en-us/azure/ai-services/content-understanding/document/analyzer-improvement states, that after adding labeled data you need to rebuild the analyzer ("After you add samples, rebuild the analyzer so the analyzer can use the samples.").

How exactly can I rebuild analyzer? I can't run "Build analyzer" with the name of existing analyzer (it show validation error "An analyzer with this name already exists."). The only way I found is to delete existing analyzer and build new one from Content Understanding Portal. There should be a more straightforward way to do that.

Azure Content Understanding in Foundry Tools
0 comments No comments

1 answer

Sort by: Most helpful
  1. AI answer

    2026-05-22T09:03:57.6+00:00

    In Content Understanding, “rebuild the analyzer” means building it again from the same project/schema after adding labeled samples; it does not require deleting and recreating the analyzer.

    Use Content Understanding Studio (Foundry portal) as follows:

    1. Open the project that defines the analyzer.
    2. Add or update labeled samples using the Label data tab on a document extraction result page, then save the corrected labels.
    3. Return to the analyzer configuration in the same project (where the schema was defined and tested).
    4. Select Build analyzer again. Provide a name (can be the same logical analyzer you are updating, as exposed via the endpoint) and confirm the build.
    5. After the build completes, use the updated analyzer from the analyzer list. The new build incorporates the labeled samples; no delete/recreate step is required beyond triggering Build analyzer.

    The key requirement is: labeled samples are added in Content Understanding Studio, and then the analyzer must be rebuilt from that project so the new examples are used at analysis time.


    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.