Редактиране

Споделяне чрез


Language support for a QnA Maker resource and knowledge bases

This article describes the language support options for QnA Maker resources and knowledge bases.

Note

The QnA Maker service is being retired on the 31st of March, 2025. A newer version of the question and answering capability is now available as part of Azure AI Language. For question answering capabilities within the Language Service, see question answering. Starting 1st October, 2022 you won’t be able to create new QnA Maker resources. For information on migrating existing QnA Maker knowledge bases to question answering, consult the migration guide.

Language for the service is selected when you create the first knowledge base in the resource. All additional knowledge bases in the resource must be in the same language.

The language determines the relevance of the results QnA Maker provides in response to user queries. The QnA Maker resource, and all the knowledge bases inside that resource, support a single language. The single language is necessary to provide the best answer results for a query.

Single language per resource

Consider the following:

  • A QnA Maker service, and all its knowledge bases, support one language only.
  • The language is explicitly set when the first knowledge base of the service is created.
  • The language is determined from the files and URLs added when the knowledge base is created.
  • The language can't be changed for any other knowledge bases in the service.
  • The language is used by the Cognitive Search service (ranker #1) and the QnA Maker service (ranker #2) to generate the best answer to a query.

Supporting multiple languages in one QnA Maker resource

This functionality is not supported in our current Generally Available (GA) stable release. Check out question answering to test out this functionality.

Supporting multiple languages in one knowledge base

If you need to support a knowledge base system, which includes several languages, you can:

  • Use the Translator service to translate a question into a single language before sending the question to your knowledge base. This allows you to focus on the quality of a single language and the quality of the alternate questions and answers.
  • Create a QnA Maker resource, and a knowledge base inside that resource, for every language. This allows you to manage separate alternate questions and answer text that is more nuanced for each language. This gives you much more flexibility but requires a much higher maintenance cost when the questions or answers change across all languages.

Languages supported

The following list contains the languages supported for a QnA Maker resource.

Language
Arabic
Armenian
Bangla
Basque
Bulgarian
Catalan
Chinese_Simplified
Chinese_Traditional
Croatian
Czech
Danish
Dutch
English
Estonian
Finnish
French
Galician
German
Greek
Gujarati
Hebrew
Hindi
Hungarian
Icelandic
Indonesian
Irish
Italian
Japanese
Kannada
Korean
Latvian
Lithuanian
Malayalam
Malay
Norwegian
Polish
Portuguese
Punjabi
Romanian
Russian
Serbian_Cyrillic
Serbian_Latin
Slovak
Slovenian
Spanish
Swedish
Tamil
Telugu
Thai
Turkish
Ukrainian
Urdu
Vietnamese

Query matching and relevance

QnA Maker depends on Azure AI Search language analyzers for providing results.

While the Azure AI Search capabilities are on par for supported languages, QnA Maker has an additional ranker that sits above the Azure search results. In this ranker model, we use some special semantic and word-based features in the following languages.

Languages with additional ranker
Chinese
Czech
Dutch
English
French
German
Hungarian
Italian
Japanese
Korean
Polish
Portuguese
Spanish
Swedish

This additional ranking is an internal working of the QnA Maker's ranker.

Next steps