The pattern to build assistants supporting languages not supported by Watson Assistant is to add a machine translation service, such as Language Translator, before interacting with WA, ensuring that such interaction is performed in a language supported by WA, regardless of the end user language.
Such pattern can be easily implemented with an application orchestrator or from the client application. But such pattern can not be implemented if user wants to leverage the channel integration connectors (Facebook, Slack and others to come). To solve this problem, WA should provide either a native integration to LT or a webhook BEFORE executing the intent and entity analysis.
Using a native integration ensures only Watson LT can be easily integrated raising usage on such service and putting Watson LT in the map. Providing a webhook opens the possibility to use other translators in the market, such as Google or Microsoft. Both options solve the problem. Which one depends on overall Watson/WA strategy.
Why is it useful?
|Who would benefit from this IDEA?||As a developer I want to support more languages than what WA can provide in an easy and integrated way.|
How should it work?