can we retrieve only the intent using watson conversation? - watson-conversation

Question

Currently we are using Watson Natural Language Classifier service (NLC) to get intent for an user's question. But configuring and maintaining NLC is becoming difficult, so was wondering whether it would be possible to only get intent of user's question using Watson Conversation section, only the intent not the response of dialog from the service.

Answer 1

The intent comes as part of the response back from conversation. If you set the parameter alternate_intents=true then the top 10 intents are returned.
You will still get the rest of the payload, but you can ignore it. I would recommend to create one dialog node with a condition of true and nothing else. This will prevent SPEL errors when not finding a matched node.
Your response will look something like this.
{
"alternate_intents": true,
"context": {
"conversation_id": "6c256e10-ba3b-4d2b-84fc-740853879d4f",
"system": {
"_node_output_map": { "True": [0] },
"branch_exited": true,
"branch_exited_reason": "completed",
"dialog_request_counter": 1,
"dialog_stack": [ { "dialog_node": "root" } ],
"dialog_turn_counter": 1
}
},
"entities": [],
"input": { "text": "test" },
"intents": [
{ "intent": "intent1", "confidence": 1.0 },
{ "intent": "intent2", "confidence": 0.9 },
{ "intent": "intent3", "confidence": 0.8 },
{ "intent": "intent4", "confidence": 0.7 },
{ "intent": "intent5", "confidence": 0.6 },
{ "intent": "intent6", "confidence": 0.5 },
{ "intent": "intent7", "confidence": 0.4 },
{ "intent": "intent8", "confidence": 0.3 },
{ "intent": "intent9", "confidence": 0.2 },
{ "intent": "intent10", "confidence": 0.1 }
],
"output": {
"log_messages": [],
"nodes_visited": [ "True" ],
"text": [ "" ]
}
}
All you need to reference is the json_response['intents']. Also if you only care about the intent, you do not need to keep sending back the context.
Just to add to this. NLC and Conversation use two very different learning models.
NLC uses "Relative Confidence"
Conversation uses "Absolute Confidence"
In the case of Relative, all confidences of items found will add up to 1. In layman terms, NLC automatically assumes that the answer can only be in the intents it has been given.
For Absolute, the confidences relate only to that intent. This means that conversation can understand that what you are saying may not be in the training it has been given. It also means that your intent list can come back empty.
So don't panic if something that was giving you 90% before is now giving you 60%. They are just scoring differently.