Index Schema: Option to have custom 'PayloadText' field (rather than including it within the Payload JSON field)

[glen-fdce]

Where a chunk of indexed text includes JSON, searchability is hindered (in Azure AI Search) since the JSON becomes escaped JSON within JSON. It would be ideal if a chunk could be represented within its own field, rather than being part of the payload field.

For example, instead of this:-

{
      "id": "ZD1ka2JiMGM5My05ODBmLTQ3aTgtYjcxOS0wODEyMThjZaRmZTMvL3A9NzgwMTI3YjE0NDdiNDA4NTa4ZTczMGUxMWQ3ZGMaOGa_",
      "tags": [
          ...
      ],
      "payload": "{\"url\":\"\",\"schema\":\"20231218A\",\"file\":\"document-name.txt\",\"text\":\"blah-blah-blah",\"vector_provider\":\"AI.AzureOpenAI.AzureOpenAITextEmbeddingGenerator\",\"vector_generator\":\"TODO\",\"last_update\":\"2025-02-27T11:37:56\"}",
      "embedding": [...]
}

This...

{
      "id": "ZD1ka2JiMGM5My05ODBmLTQ3aTgtYjcxOS0wODEyMThjZaRmZTMvL3A9NzgwMTI3YjE0NDdiNDA4NTa4ZTczMGUxMWQ3ZGMaOGa_",
      "tags": [
          ...
      ],
      "payloadText": "blah-blah-blah",
      "payload": "{\"url\":\"\",\"schema\":\"20231218A\",\"file\":\"document-name.txt\",\"vector_provider\":\"AI.AzureOpenAI.AzureOpenAITextEmbeddingGenerator\",\"vector_generator\":\"TODO\",\"last_update\":\"2025-02-27T11:37:56\"}",
      "embedding": [...]
}

[dluc]

I'd recommend looking into Semantic Kernel Microsoft Extensions Vector Data library, that allows to create and query custom schemas.