Dialogflow.com is a chatbot / AI platform that allows to easily train a bot in order to detect the user intent and extract relevant data.

For a complete tutorial about Dialogflow.com give a look here. The Dialogflow.com node sends the text of the incoming message to the Dialogflow.com servers, then:

• if no intent is detected, then the incoming message is redirected to the second output
• if an intent is detected, then it’s stored in the context variable topic (or a different variable) and a key/value hash of all relevant information extracted is sent to the first output along with other informations.

Dialogflow.com is able to create a full featured chatbot, nevertheless it can be used just for the NLP capabilities and to detetect/classify user’s sentences.

For example, create a Drinks entity with these values

martini rosso
mojito
moscow mule
old fashion

Then create an intent want-drink and train with these phrases

I want a martini rosso
May I have a martini
Can I have a moscow mule
May I have a mojito

and verify that the words “martini rosso”, “martini”, “moscow mule” and “mojito” are already marked with the entity Drinks, if not, it’s possible to mark portions of the sentences with the appropriate entity. Then prepare a flow like this

If the user says “I want a martini”, the topic will be switched to “want-drink” and the output will be

{
  payload: {
    intent: 'want-drink',
    answer: 'I am the answer',
    variables: {
      Drinks: 'martini rosso'
    }
}