Azure AI service - Translation - Dictionary lookup is missing a translation

Hi @Chris Aeberhard ,

Thank you for reaching out to Microsoft Q&A.

I understand that you are experiencing issues with the dictionary lookup feature when translating the English word "mine" to French, and you would like to ensure you get the correct translation using part of speech information. I will be happy to assist you with this.

To address this, you have two possible approaches:

1. ****Direct Translation: **** You can use the Translator Text API directly to translate the word "mine" to French. The Translator Text API provides a simple interface for translating text and is not dependent on the dictionary lookup feature. By specifying the source and target languages, as well as any additional translation options, you can obtain the translation you desire. You can find more information about using the Translator Text API in the Azure documentation here
2. Part of Speech (POS) Tagging: Yes, you can use the POS tagging feature of the Translator Text API to disambiguate between different meanings of the same word. The POS tagging feature identifies the grammatical function of each word in a sentence, such as whether it is a noun, verb, adjective, etc. This information can help you determine the correct translation of a word based on its context.

Below is an example Python code snippet that demonstrates how to use the Translator Text API to translate a sentence and get the POS tags for each word. The output of this code will include the translation and POS tags, which indicate the grammatical function of each word in the translated text.

import os, requests, uuid, json
import spacy

# Initialize the spacy NLP model for English
nlp = spacy.load("en_core_web_sm")

subscription_key = 'AZURE_TEXT_TRANSLATION_API_KEY'
endpoint = 'AZURE_TEXT_TRANSLATION_ENDPOINT'
path = '/translate?api-version=3.0'
params = '&to=fr'
constructed_url = endpoint + path + params

headers = {
    'Ocp-Apim-Subscription-Key': subscription_key,
    'Content-type': 'application/json',
    'X-ClientTraceId': str(uuid.uuid4())
}

# List of texts to translate and tag
texts = [
    'iron mine',
    'This book is mine',
    'Diamond mine',
    'Coal mine',
    'iron mine'
]

for text in texts:
    body = [{
        'text': text
    }]

    response = requests.post(constructed_url, headers=headers, json=body)
    response_json = json.loads(response.content.decode('utf-8'))

    # Get the translation
    translation = response_json[0]['translations'][0]['text']

    # Process the translated text with spacy for POS tagging
    translated_doc = nlp(translation)

    # Extract POS tags for each token in the translated text
    translated_pos_tags = [token.pos_ for token in translated_doc]

    print('Original Text:', text)
    print('Translation:', translation)
    print('POS tags for translated text:', translated_pos_tags)
    print()

Output :

Original Text: iron mine
Translation: mine de fer
POS tags for translated text: ['PROPN', 'X', 'PROPN']

Original Text: This book is mine
Translation: Ce livre est le mien
POS tags for translated text: ['NOUN', 'NOUN', 'X', 'X', 'X']

Original Text: Diamond mine
Translation: Mine de diamants
POS tags for translated text: ['PROPN', 'X', 'NOUN']

Original Text: Coal mine
Translation: Houillère
POS tags for translated text: ['NOUN']

Original Text: iron mine
Translation: mine de fer
POS tags for translated text: ['PROPN', 'X', 'PROPN']

If you choose the first approach, you won't rely on the dictionary lookup and can obtain translations directly. If you choose the second approach, you can leverage POS tagging to make sure you are getting the correct translation based on the context.

If you would still like to provide feedback regarding the dictionary lookup feature, you can submit your feedback through the Azure Feedback Forum here. This forum is monitored by the Azure product team, and they use the feedback provided by customers to improve the service.

Hope this information helps!