Dialog is not following stories.md after introducing custom actions

Rasa Open Source
1

Rasa Core version :0.12.3

Python version : Python 3.6.8

Operating system (windows, osx, …): Ubuntu 16.04

Issue : dialog is not following stories.md after introducing custom actions. Using Jupyter notebook for writing and testing the code.

I am really new to Rasa and have been exploring it for the purpose of building a chatbot. Having a difficult time when it comes to custom actions.

I was playing with utterances in the actions till now and the flow of dialog was going fine. I tried writing some custom actions for the chatbot but the flow gets struck in the first conversation itself.

a path in stories.md file is as below:-

#happy path model enquiry 1

  • greet
    • utter_greet
  • Customer_enquiry_model
    • action_get_model_config
    • utter_did_that_help
  • mood_affirm
    • utter_mood_affirm
  • OPPO_Phones_Price
    • action_get_model_price
    • utter_did_that_help
  • mood_affirm
    • utter_mood_affirm
  • goodbye
    • utter_goodbye

For the path mentioned , I am getting :-

Your bot is ready to talk! Type your messages here or send ‘stop’

Me :Hi I am Arghya

Bot : I am not sure what you are aiming for.

expected :- utter_greet

it seems that the bot is not able to predict the action as in the path.

Please guide me where i am going wrong and nudge me to the right direction.

the nlu.md training file is as below :-

intent:CEO_info

intent:CustomerCare_info

intent:Customer_enquiry_model

  • Yes I am looking for f10
  • I want to know about f20
  • tell me about f10
  • give me the detail of model name
  • I would say model name
  • details about f20
  • information about f30
  • enlighten me about x10
  • looking for this model

intent:OPPO_Phones_Price

  • want to buy F7 what is the price?
  • wanna buy E10, tell me the price?
  • interested in f7, give me the price
  • interested to go for t66, price pls
  • what is the market value of f7?
  • may I know the worth of f7?
  • can I have the price of g14?
  • let me know the price of e10?
  • share the price of x20?
  • want to know the price of i10?
  • price of ur phone please?
  • do u have the price of y10
  • price please
  • what is the price
  • price of this particular phone

intent:Official_Website

intent:Oppo_store_ocation

  • shop address please
  • Where is your shop situated?
  • your nearest shop address
  • tell me your location please
  • where are you located at?
  • give me your shop address
  • where is your outlet at?
  • nearest store please
  • where is the nearest store?
  • where is the nearest store of oppo?
  • shop near to me
  • address of the shop near to me
  • location of the shop near to me

intent:Oppo_product_comparison

intent:Service_Center_OR_Factory

intent:about_OPPO

  • ollie what do you know about oppo?
  • what is oppo?
  • what do you know about oppo?
  • which field you belong to?
  • what is ur business?
  • what is oppo about?
  • what do u do?
  • can i know about ur services?
  • what does Oppo do?

intent:brand_info

  • Ollie what brand you belong to?
  • what is ur brand?
  • brand name?
  • which brand do you represent?
  • which brand do you work for?
  • give me some information about the brand you are working for.
  • may I know about your brand name?
  • which brand are you?

intent:goodbye

  • see you
  • good by
  • see you later
  • bye
  • goodbye
  • have a nice day
  • see you around
  • bye bye
  • see you later
  • that’s all!Thank you.
  • Nope.take care.
  • I got all I need.Thanks.
  • It was really helpful.Thanks.
  • Ciao

intent:greet

  • hey I am Peter
  • hello there I am Hans
  • hi I am Tom
  • hello there
  • good morning
  • good evening
  • morning
  • hey there
  • let’s go
  • hey dude
  • good morning
  • good evening
  • good afternoon
  • Hi,I am Arghya
  • hi
  • hello
  • hi Arghya here

intent:city_name

intent:model_name_lookout

  • I want to know about g12 and x12
  • tell me about t16 and f7
  • I am interested in Model1 and Model2.
  • give me the detail of Model1 and Model2
  • Take Model1 and Model2
  • I would say Model1 and Model2
  • compare Model1 and Model2

intent:mood_affirm

  • yes
  • indeed
  • of course
  • that sounds good
  • correct
  • sure
  • exactly
  • hundred percent correct
  • that’s right
  • yup
  • yeep
  • sounds right
  • looks nice

intent:mood_deny

  • no
  • never
  • I don’t think so
  • don’t like that
  • no way
  • not really
  • nope
  • na
  • not exactly
  • double no
  • not that

intent:provide_founder

  • Oppo is Owned by whom?
  • Oppo is established by?
  • Can u tell me owner
  • who is oppo’s owner
  • oppo’s owner
  • who owns Oppo?
  • Oppo founded by whom?
  • Who is the founder of oppo?
  • oppo is owned by whom?
  • oppo mobile is owned by whom?
  • who owns oppo?
  • tell me the name of the owner of oppo.
  • name of the owner?
  • owner’s name?
  • owner please?
  • who is the owner of oppo?

lookup:PERSON

/home/arghya/rasa_test/data/place_finder/PERSON.txt

lookup:model

/home/arghya/rasa_test/data/place_finder/model.txt

lookup:customer_location

/home/arghya/rasa_test/data/place_finder/customer_location.txt

action server code :-

from future import absolute_import

from future import division

from future import unicode_literals

from rasa_core.domain import Domain

from rasa_core.trackers import EventVerbosity

import logging

logger = logging.getLogger( name )

import requests

import json

from rasa_core_sdk import Action

from rasa_core_sdk.events import SlotSet

from rasa_core_sdk.events import UserUtteranceReverted

from rasa_core_sdk.events import AllSlotsReset

from rasa_core_sdk.events import Restarted

class SaveCityName(Action):

 def name(self):
     return 'action_save_city_name'

 def run(self, dispatcher, tracker, domain):
    orig = next(tracker.get_latest_entity_values("customer_location"), None)
    if not orig:
        dispatcher.utter_message("Please enter a valid city name")
        return [UserUtteranceReverted()]
    return [SlotSet('customer_location',orig)]

class SaveModelName(Action):

def name(self):
     return 'action_save_model_name'
    
def run(self, dispatcher, tracker, domain):
    inp = next(tracker.get_latest_entity_values("model"), None)
    if not inp:
        dispatcher.utter_message("Please enter a valid model")
        return [UserUtteranceReverted()]
    return [SlotSet('model',inp)]

class ActionSlotReset(Action):

def name(self): 
    return 'action_slot_reset'
def run(self, dispatcher, tracker, domain): 
    return[AllSlotsReset()]

class getServiceCentreLocation(Action):

def name(self):
    return 'action_get_service_centre_location'
def run(self):
    print("you will receive the service location once the database is up.Working on it :) ")
    pass

class getShopLocation(Action):

   def name(self):
       return 'action_get_shop_location'
  def run(self):
    print("you will receive the shop location once the database is up and running.Thanks for the wait.Next Question please? :)")
    pass

class getModelPrice(Action):

def name(self):
    return 'action_get_model_price'
def run(self):
    print("you will receive the price of the model once the database is up and running.Thanks for the wait.Next Question please? :)")
    pass

class getModelConfig(Action):

def name(self):
    return 'action_get_model_config'
def run(self):
    print("you will receive the model configuration once the database is up and running.Thanks for the wait.Next Question please? :)")
    pass

model.txt PERSON.txt

Content of domain file (if used & relevant): domain.yml

intents:

  • greet
  • goodbye
  • provide_founder
  • about_OPPO
  • brand_info
  • CustomerCare_info
  • Service_Center_OR_Factory
  • Oppo_store_location
  • CEO_info
  • Official_Website
  • OPPO_Phones_Price
  • Customer_enquiry_model
  • Oppo_product_comparison
  • model_name_lookout
  • mood_affirm
  • mood_deny
  • city_name

slots: PERSON: type: text model: type: text customer_location: type: text workshop_location: type: text shop_location: type: text

entities:

  • PERSON
  • information
  • customer_location
  • workshop_location
  • shop_location
  • web_address
  • model
  • feature

actions:

  • utter_greet

  • utter_goodbye

  • utter_did_that_help

  • utter_unclear

  • utter_city_name

  • utter_Oppo_shop_location

  • utter_provide_founder

  • utter_about_OPPO

  • utter_brand_info

  • utter_CustomerCare_info

  • utter_Service_Center_OR_Factory

  • utter_CEO_info

  • utter_Official_Website

  • utter_OPPO_Phones_Price

  • utter_Customer_enquiry_model

  • utter_Oppo_product_comparison

  • utter_model_name_lookout

  • utter_mood_affirm

  • utter_mood_deny

  • utter_happy

  • utter_more_info

  • utter_thanks

  • utter_deny

  • utter_confirm

  • utter_check_another_one

  • action_save_city_name

  • action_get_service_centre_location

  • action_slot_reset

  • action_get_shop_location

  • action_save_model_name

  • action_get_model_price

  • action_restart

  • action_get_model_config templates: utter_greet:

    • text: “Hey, {PERSON}!This is Ollie,your assistant.How may I help you?”
    • text: “Hi {PERSON}!I am here to help.Shoot!”
    • text: “Hi,how can I help you?”

    utter_goodbye:

    • text: “Bye {PERSON} and take care!”
    • text: “Bye”

    utter_did_that_help:

    • text: “Did that help you {PERSON}?”

    utter_unclear:

    • text: “I am not sure what you are aiming for.”

    utter_thanks:

    • text: “You are very welcome.”
    • text: “Glad I could help!” utter_deny:
    • text: “That’s a shame. Let me know if you change your mind.”

    utter_provide_founder:

    • text: “Oppo’s founder is Chen Mingyong.”

    utter_about_OPPO:

    • text: “Oppo Electronics Corporation, commonly referred to as Oppo, is a Chinese consumer electronics and mobile communication company, known for its smartphones, Blu-ray players and other electronic devices. A leading manufacturer of smartphones, Oppo was the top smartphone brand in China in 2016 and was ranked No. 8 worldwide”

    utter_brand_info:

    • text: “I work for Oppo Mobile”

    utter_CustomerCare_info:

    • text: “Oppo’s customer care number is 1800 103 2777”

    utter_CEO_info:

    • text: “CEO of Oppo is Chen Mingyong.Started his position from 2/11/2017.”

    utter_Official_Website:

    • text: “Official website is: www.oppo.in”

    utter_Oppo_product_comparison:

    • text: “Decision is yours:Here is the feature list of {model} and {model}:”

    utter_happy:

    • text: “Great carry on!”

    utter_more_info:

    • text: “Please provide more details so that i can understand your query”

    utter_mood_affirm:

    • text: “Glad that i was useful.Next Question if any?”

    utter_mood_deny:

    • text: “Try again!”

    utter_city_name:

    • text: “Which city are you from {PERSON}?”
    • text: “Which city are you from?”
    • text: “Name of the city please?”

    utter_confirm:

    • text: “I will be making inquiry for customer_location: {customer_location}. Is that correct?” utter_check_another_one:
    • text: “Do you want to make another inquiry?”
2

This basically means your intent isn’t getting classified correctly. Run the NLU server and hit the chat query once to see if it correctly identifies the intent

3

I checked the chat query and the intents are being classified correctly by the NLU server. for the query : “Hi I am Arghya” the result of pprint(interpreter.parse(“I am Arghya”)) is

{‘entities’: [{‘confidence’: None, ‘end’: 14, ‘entity’: ‘PERSON’, ‘extractor’: ‘ner_mitie’, ‘start’: 8, ‘value’: ‘Arghya’}], ‘intent’: {‘confidence’: 0.6571330642134279, ‘name’: ‘greet’}, ‘intent_ranking’: [{‘confidence’: 0.6571330642134279, ‘name’: ‘greet’}, {‘confidence’: 0.09446779527877744, ‘name’: ‘city_name’}, {‘confidence’: 0.0400228170128286, ‘name’: ‘mood_deny’}, {‘confidence’: 0.03340266453139138, ‘name’: ‘goodbye’}, {‘confidence’: 0.019552058361452504, ‘name’: ‘Customer_enquiry_model’}, {‘confidence’: 0.019350730188458353, ‘name’: ‘Oppo_product_comparison’}, {‘confidence’: 0.01843151750064483, ‘name’: ‘mood_affirm’}, {‘confidence’: 0.016625951193141118, ‘name’: ‘CustomerCare_info’}, {‘confidence’: 0.016560706697613795, ‘name’: ‘model_name_lookout’}, {‘confidence’: 0.013907684670814468, ‘name’: ‘provide_founder’}], ‘text’: ‘Hi I am Arghya’}