Hey
Thanks for suggestions.
I don’t see how to dynamically create the question to the user with FormAction. I go through an action that sends a random question to the user.
Config
language: "fr"
pipeline:
- name: WhitespaceTokenizer
case_sensitive: False
- name: RegexFeaturizer
- name: LexicalSyntacticFeaturizer
- name: CountVectorsFeaturizer
analyzer: "char_wb"
min_ngram: 1
max_ngram: 4
- name: EntitySynonymMapper
- name: ResponseSelector
epochs: 100
- name: CRFEntityExtractor
BILOU_flag: true
- name: "DucklingHTTPExtractor"
url: "http://localhost:8000"
dimensions: ["time", "number", "amount-of-money", "distance"]
locale: "fr_FR"
timezone: "Europe/Paris"
timeout : 3
- name: DIETClassifier
entity_recognition: false
epochs: 100
policies:
- name: AugmentedMemoizationPolicy
max_history: 15
- name: TEDPolicy
max_history: 5
epochs: 100
- name: FormPolicy
- name: MappingPolicy
- name: FallbackPolicy
nlu_threshold: 0.1
ambiguity_threshold: 0.04
core_threshold: 0.3
fallback_action_name: utter_oupsomethingfailed
NLU
# intent:choice_category
[aleatoire](category_questions)
[caractere](category_questions)
[couple](category_questions)
[sexe](category_questions)
[loisirs](category_questions)
[cinema](category_questions)
[humour](category_questions)
[quotidien](category_questions)
## intent:answer_question
[AC13_1](answer_id)
## regex:answer_id
- [A-Z]{1,3}[0-9]{1,4}_[0-9]{1,4}
Domain
actions:
- action_menu
- action_ask_question
- utter_ask_criteries_genre
- utter_ask_genre
- utter_ask_mood
- utter_goodbye
- utter_greet
- utter_happy
- utter_iamabot
- utter_menu
- utter_oupsomethingfailed
- utter_thanks
- utter_unhappy
- utter_validate_criteries_genre
entities:
- criteries_sexe
- sexe
- module
- category_questions
- question_id
- answer_id
- answer_question
intents:
- deny
- thanks
- greet
- sexe
- mood_great
- affirm
- mood_unhappy
- goodbye
- bot_challenge
- module
- choice_category
- menu
slots:
module:
type: unfeaturized
category_questions:
type: unfeaturized
question_id:
type: unfeaturized
answer_id:
type: unfeaturized
criteries_sexe:
type: categorical
values:
- homme
- femme
- bi
- __other__
sexe:
type: categorical
values:
- homme
- femme
- autre
- __other__
responses:
utter_ask_criteries_genre:
- buttons:
- payload: /sexe{"criteries_sexe":"homme"}
title: 👨🦱 Homme
- payload: '/sexe{"criteries_sexe": "femme"}'
title: 👩 Femme
- payload: '/sexe{"criteries_sexe": "bi"}'
title: Bi
text: Et que recherches-tu ?
utter_ask_genre:
- buttons:
- payload: /sexe{"sexe":"homme"}
title: 👨🦱 Homme
- payload: '/sexe{"sexe": "femme"}'
title: 👩 Femme
- payload: '/sexe{"sexe": "autre"}'
title: Non genré
text: De quel genre te désignes-tu ?
utter_ask_mood:
- text: Comment vas-tu aujourd'hui ?
utter_goodbye:
- text: Bye
- text: Prends soin de toi, à bientôt 👋
- text: Bye 👋
- text: A plus tard 👋
- text: Au plaisir de te reparler 👋
utter_greet:
- text: Salut, moi c'est Archelot 🤖
- text: Bonjour! 😇 Archelot! Ravi de te rencontrer.
utter_happy:
- text: Great, carry on!
utter_iamabot:
- text: Je suis Archelot, une IA imaginé par mes développeurs Valentin et Coralie.
utter_menu:
- buttons:
- payload: /menu{"module":"notifications"}
title: News
- payload: /menu{"module":"questions"}
title: Parler
- payload: /menu{"module":"help"}
title: FAQ
text: Que souhaites-tu faire?
utter_oupsomethingfailed:
- text: Oups, je n'ai pas compris ta réponse... Après tout je ne suis qu'un bot
🤖
utter_thanks:
- text: Au plaisir 😇
utter_unhappy:
- text: Ouais c'est la vie !
utter_validate_criteries_genre:
- text: 'Merci 😇 Genre : {sexe} et recherche : {criteries_sexe}'
session_config:
carry_over_slots_to_new_session: false
session_expiration_time: 60
Story
## répondre à des questions caractère
* menu{"module": "questions"}
- action_menu
* choice_category{"category_questions": "caractere"}
- slot{"category_questions": "caractere"}
- action_ask_question
## répondre à des couple
* menu{"module": "questions"}
- action_menu
* choice_category{"category_questions": "couple"}
- slot{"category_questions": "couple"}
- action_ask_question
actions.py
from datetime import datetime
from typing import Any, Dict, List, Text, Union, Optional
import json
import logging
from rasa_sdk import ActionExecutionRejection
from rasa_sdk import Action, Tracker
from rasa_sdk.events import EventType, SlotSet, FollowupAction, UserUtteranceReverted
from rasa_sdk.executor import CollectingDispatcher
from rasa_sdk.forms import FormAction, REQUESTED_SLOT, Action
from rasa.core.slots import Slot
from typing import Dict, Text, Any, List, Union
import random
import re
import sqlite3
import requests
logging.basicConfig(filename='example.log',level=logging.DEBUG)
logging.debug('This message should go to the log file')
logging.info('So should this')
logging.warning('And this, too')
cateogory =[
{"category" : "Santé", "slug" : '/choice_category{"category_questions": "sante"}'},
{"category" : "Public", "slug" : '/choice_category{"category_questions": "public"}'},
]
db_questions = [
{"a" : "public",
"questions" : [
{
"id" : 1,
"message" : "Tu vas à la plage cet été",
"buttons" : [
{"answer" : "Je pense oui", "slug" : '/answer_question{"answer_id": "AC01_1"}'},
{"answer" : "Pas les moyens", "slug" : '/answer_question{"answer_id": "AC01_2"}'},
]
}
]
class menu_action(Action):
def name(self):
return "action_menu"
def run(self, dispatcher, tracker, domain) -> List[EventType]:
buttons = []
if tracker.get_slot('module') == 'questions':
message_title = "De quoi veux-tu discuter?"
for i in range(len(cateogory[0])) :
buttons.append(
{
"title": cateogory[i]["category"],
"payload": cateogory[i]["slug"],
}
)
dispatcher.utter_message(message_title, buttons=buttons)
return []
class action_ask_question(Action):
def name(self) -> Text:
return "action_ask_question"
def run(self, dispatcher: CollectingDispatcher,
tracker: Tracker,
domain: Dict[Text, Any]) -> List[Dict[Text, Any]]:
id_question = 0
message_title = "?"
if tracker.get_slot('category_questions') :
slot_category = tracker.get_slot('category_questions')
buttons = []
for categories in db_questions:
if categories['a'] == slot_category :
choice = random.choice(categories['questions'])
id_question = choice['id']
message_title = choice['message']
for answer in choice['buttons']:
buttons.append(
{
"title": answer['answer'],
"payload": answer['slug'],
}
)
#Ici je renvoi la question et les réponses possible à afficher
dispatcher.utter_message(message_title, buttons=buttons, tracker=tracker)
SlotSet("question_id", id_question)
return []
Theoretically, I have to go through an action for the answer but this intent cannot contain all possible answers. Is it possible with a FormAction to retrieve them automatically?
rasa interactive
5 slot{"category_questions": "public"}
action_ask_question 1.00
Question to ask ?
Buttons:
1: answer (/answer_question{"answer_id": "AC02_1"})
2:
answer 2 (/answer_question{"answer_id": "AC02_2"})
Type out your own message...
action_listen 1.00
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
6 /answer_question{"answer_id": "AC02_2"}
intent: answer_question 1.00
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
7 slot{"answer_id": "AC02_2"}
action_ask_question
Do you prefer ?
Buttons:
1: Answer 1 (/answer_question{"answer_id": "AC03_1"})
2:
Answer 2 (/answer_question{"answer_id": "AC03_2"})
Type out your own message...
action_listen 0.99
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
8 /answer_question{"answer_id": "AC03_1"}
intent: answer_question 1.00
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
9 slot{"answer_id": "AC03_1"}
utter_menu
Do I have to restart an action when the user has replied in order to retrieve his answer?
What do you think about doing that?
