'Azure cognitive service language detection reports Spanish for input "no"

Submitting a request to the language detection API with input text no

curl --location --request POST 'https://api.cognitive.microsofttranslator.com/detect?api-version=3.0
' \
--header 'Ocp-Apim-Subscription-Key: XXXXXXXXXXXXXXXXXXXX' \
--header 'Content-Type: application/json' \
--data-raw '[
    { "Text": "no" }
]'

Returns the following response:

[
    {
        "language": "es",
        "score": 1.0,
        "isTranslationSupported": true,
        "isTransliterationSupported": false
    }
]

Other services detect that there is some nuance here:

{
  "data": {
    "detections": [
      {
        "language": "en",
        "isReliable": true,
        "confidence": 1.6
      },
      {
        "language": "es",
        "isReliable": false,
        "confidence": 1.6
      },
      {
        "language": "pt",
        "isReliable": false,
        "confidence": 1.6
      }
    ]
  }
}

Is there any way to modify the request, or to create a custom translator resource which can return multiple detected languages and allow the client to choose?

azureazure-cognitive-serviceslanguage-detection


Sources

This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.

Source: Stack Overflow

Solution Source

Related Questions