WARNING rasa_core.training.dsl - Found unknown intent 'inform

Rasa Open Source
1

I am trying the bot with rasa. I have some issues as follows: intents:

  • greet
  • affirm
  • deny
  • inform
  • thankyou
  • request_info
  • booking entities:
  • location
  • info
  • people
  • price
  • cuisine
  • road

I have some sample stories in my story file as follows:

  • greet
    • utter_default
    • utter_ask_cuisine
  • inform(‘cuisine’: ‘\u5ddd\u83dc’)
    • utter_ask_location
  • inform(‘location’: ‘\u4e0a\u6d77\u5e02’)
    • utter_ask_road
  • inform(‘road’: ‘\u53e4\u7f8a\u8def’)
    • utter_ask_numpeople
  • inform(‘people’: ‘2\u4e2a’)
    • utter_ask_price
  • inform(‘price’: ‘\u4fbf\u5b9c\u7684’)
    • utter_ask_moreupdates
  • affirm
    • action_search_restaurants
  • thankyou
    • utter_goodbye
  • out_of_scope
    • utter_default
  • deny
    • utter_ack_findalternatives
    • action_suggest
  • thankyou
    • utter_goodbye

I got the warning messages as follows when I train core: WARNING rasa_core.training.dsl - Found unknown intent ‘inform(“price”: “u4fbfu5b9cu7684”)’ on line 8351. Please, make sure that all intents are listed in your domain yaml. Could you any guys help me to solve the issue? Thanks.

Juven

2

Try the correct format, your parenthesis here should be curly brackets, i.e. inform{'road': '\u53e4\u7f8a\u8def'}

Also what version are you on? You should be able to put unicode characters directly into your stories.

3

rasa-nlu=0.14.4 rasa-core=0.13.2 rasa-core-sdk=0.12.1 language=Chinese encodings in story = utf-8

4

Okay. I believe that you can write the characters directly into the stories with no problem, but know there used to be a bug with interactive learning where it would output in that format. This is fixed in rasa 1.1. Do the curly brackets work for you, though?

5

I have upates parenthesis in my story, but I still see the warning as follows: 2019-06-14 16:05:19 WARNING rasa_core.training.dsl - Found unknown intent ‘out_of_scope’ on line 13870. Please, make sure that all intents are listed in your domain yaml. 2019-06-14 16:05:19 WARNING rasa_core.training.dsl - Found unknown intent ‘inform(‘location’: ‘\u5317\u4eac\u5e02’)’ on line 13881. Please, make sure that all intents are listed in your domain yaml. 2019-06-14 16:05:19 WARNING rasa_core.training.dsl - Found unknown intent ‘inform(‘road’: ‘\u53e4\u7f8a\u8def’)’ on line 13883. Please, make sure that all intents are listed in your domain yaml.

And give me another warning: Found unknown intent ‘out_of_scope’ on line 13870, what meaning with this?

Thanks.

6

If you look at the story you posted, it contains an out_of_scope intent that is not in your domain.

Also, from the error, it looks like you haven’t updated all of the parenthesis, or you haven’t retrained since you did.

7

I did as you said to update parenthesis in my story file.I got the new warning messges as follows:

2019-06-14 16:28:39 WARNING rasa_core.interpreter - Invalid to parse arguments in line ‘inform{‘location’: ‘\u5317\u4eac\u5e02’}’. Failed to decode parametersas a json object. Make sure the intentfollowed by a proper json object. Error: Expecting property name enclosed in double quotes: line 1 column 2 (char 1)

8

I’s ok after I have upates all simgle quotes with double quotes. Thanks.

9

An another common questions is:

Could you give me some advices or principles about NLU training data?

10

What sort of advice are you looking for?

11

For examples: I have five entities:

  • location
  • people
  • price
  • cuisine
  • road and the following intents:
  • greet
  • affirm
  • deny
  • inform
  • thankyou
  • request_info
  • booking

intents:

  • greet
  • affirm
  • deny
  • inform
  • thankyou
  • request_info
  • booking My question:whether do I need to prepare seperative examples one by one or combinatorial format in trainning data,such as(in chatito format): seperative examples one by one: I want to book restaurant in {Location}. I want to book restaurant for {people} people. I want to book restaurant in round {road}. I want to book {price} restaurant. I want to book {cuisine} restaurant. or use the combinatorial format: I want to book {price?} {cuisine?} restaurant for {people?} people in round {road?} {location?} Could give more useful advice how to choice? which more effective for NLU training? Thanks.
12

I am facing same issue in rasa ==1.6.0 when adding new intent.plz guide me