From reading this http://rasa.com/docs/rasa/core/responses/
it seems that getting responses from the core can be either done by entering them in the domain.yml
file or external CMS. Is it possible to query the responses from a local SQLite database?
Hi @wassimseif, there is no out-of-the box support to query an SQLite database to do that. You could override the generate
method of the TemplatedNaturalLanguageGenerator
and include your SQLite retrieval there. That method takes the template name as one of its arguments, which you can use to retrieve the desired template. You would then start the nlg server with python examples/nlg_server/nlg_server.py -p PORT -d DOMAIN
. You’ll have to specify this new server in your endpoints.yml
:
nlg:
url: http://localhost:5055/nlg # url of the nlg endpoint
Hi,
Is there a small sample of custom NLG
server sending out responses
I am confused on how will the responses (which assistant sends back to the user) be generated outside of Rasa Core ? Will the custom NLG
server parse the incoming /POST request (text and intent_ranking) & respond back with a message based on intent ?