Botframework-solutions: spanish Language translation / No answer from cognitive models

Created on 18 Jun 2019  路  3Comments  路  Source: microsoft/botframework-solutions

Project

Name: Virtial Assistant

Language: Typescript

Description

Virtual Assistanthot works for languages different than english.

To Reproduce

I've made a brand new project using the yeoman's generator-botbuilder-assistant.
Then i've executed the bot locally using appId and password in blank, then installing node dependencies and after that starting the bot.
Then i use the emulator to test the bot. By default, the language of the emulator is "en-US".
When i test the generated bot, it shows the onboarding dialog and the links to test luis and qna works well. However when i try to switch the language to "es" or "es-CO", the onboarding dialog is shown in spanish but nothing else happens: utterance sent to the model seems not being processed and qna or luis didn't answer the questions.

Expected behavior

that bot could answer the questions in spanish

Screenshots

  • spanish (es-CO) nothing else seems to work after configure locale
    2
  • spanish (es-CO) onboarding dialog: the one and only dialog that works with "es" locale
    1

  • english (en-US) model works like a charm
    3

Additional context

I don't know if i must deploy another model to luis, qna or dispatcher to have the expected behavior (but this step is not commented in the virtual assistant instructions).

dependencies are:
"dependencies": { "@microsoft/microsoft-graph-client": "^1.3.0", "@microsoft/microsoft-graph-types": "^1.5.0", "azure-cognitiveservices-contentmoderator": "^4.0.0", "botbuilder": "^4.4.0", "botbuilder-ai": "^4.4.0", "botbuilder-applicationinsights": "^4.4.0", "botbuilder-azure": "^4.4.0", "botbuilder-core": "^4.4.0", "botbuilder-dialogs": "^4.4.0", "botbuilder-skills": "^4.4.6", "botbuilder-solutions": "^4.4.6", "botframework-config": "^4.4.0", "botframework-connector": "^4.4.0", "botframework-schema": "^4.4.0", "dotenv": "^6.0.0", "i18next": "^15.0.6", "i18next-node-fs-backend": "^2.1.1", "ms-rest-azure": "^2.5.0", "restify": "^7.2.1" },

Typescript Bug

Most helpful comment

Hi @andhdo,

First of all, thanks for the level of detail in the issue!

By default, the deploy.ps1 script takes en-us as the language for the cognitive models, so, if you want to have another language or more than one, you should pass the language or languages to the -languages argument.

The following example deploys the cognitive models in en-us and es-es languages

pwsh.exe -ExecutionPolicy Bypass -File deploymentscriptsdeploy.ps1 -name "" -location "" -appId "" -appPassword "" -luisAuthoringKey "" -luisAuthoringRegion "" -languages "es-es,en-us"

If you would like to further customize your deployment, check Deploy your bot which is included in Deploy your assistant documentation.

All 3 comments

Hi @andhdo,

First of all, thanks for the level of detail in the issue!

By default, the deploy.ps1 script takes en-us as the language for the cognitive models, so, if you want to have another language or more than one, you should pass the language or languages to the -languages argument.

The following example deploys the cognitive models in en-us and es-es languages

pwsh.exe -ExecutionPolicy Bypass -File deploymentscriptsdeploy.ps1 -name "" -location "" -appId "" -appPassword "" -luisAuthoringKey "" -luisAuthoringRegion "" -languages "es-es,en-us"

If you would like to further customize your deployment, check Deploy your bot which is included in Deploy your assistant documentation.

Hi @Batta32 ; it works now, generating the spanish model to the infrastructure as a service code, so both knowledge base of qna model and luis intents were published in azure.

we should bear in mind that locales depends on the list supported by each cognitive model. Currently "es-es" and "es-mx" are valid parameters.

Thanks for your help.

Hi @andhdo, it was a pleasure to help you.

We will close the issue because you could make it work! Feel free to reopen if you have a similar issue.

Was this page helpful?
0 / 5 - 0 ratings