Hello,
First sorry for my english (I’m french 
).
I have the latest configuration (Rasa Version:2.8.4) and I don’t understand how synonym and entities extraction works with my configuration in the pipeline. I read all of the documentation, testing differents configurations but I don’t have the result expected.
Here, are my differents files (I post only the config and the others files are attached because they are too long) :
config.yml
# Configuration for Rasa NLU.
# https://rasa.com/docs/rasa/nlu/components/
language: fr
pipeline:
- name: SpacyNLP
model: fr_core_news_lg
case_sensitive: False
- name: SpacyTokenizer
intent_tokenization_flag: False
- name: SpacyFeaturizer
- name: RegexFeaturizer
case_sensitive: False
- name: LexicalSyntacticFeaturizer
features: [
[ "low", "title", "upper" ],
[ "BOS", "EOS", "low", "upper", "title", "digit", "pos"],
[ "low", "title", "upper" ],
]
- name: CountVectorsFeaturizer
- name: CountVectorsFeaturizer
analyzer: "char_wb"
min_ngram: 1
max_ngram: 4
use_shared_vocab: False
- name: SpacyEntityExtractor
dimensions: ["PER"] #https://miro.medium.com/max/700/1*lJ1hSNVyCNGMG-cPiCeU1g.png //"LOC", "ORG", "MISC"
- name: EntitySynonymMapper
- name: CRFEntityExtractor
BILOU_flag: True
features: [
["low", "title", "upper"],
["bias", "low", "prefix5", "prefix2", "suffix5",
"suffix3", "suffix2", "upper", "title", "digit",
"pattern", "pos", "pos2"],
["low", "title", "upper"]
]
featurizers: []
split_entities_by_comma:
special_type: False
any_information: False
- name: EntitySynonymMapper
- name: RegexEntityExtractor
case_sensitive: False
use_lookup_tables: True
use_regexes: False
- name: EntitySynonymMapper
- name: SklearnIntentClassifier
- name: ResponseSelector
epochs: 100
retrieval_intent: chitchat
constrain_similarities: true
model_confidence: softmax
- name: FallbackClassifier
threshold: 0.3
ambiguity_threshold: 0.01
- name: "DucklingEntityExtractor"
url: "http://172.17.0.2:8000/"
dimensions: ["distance", "duration", "numeral", "ordinal", "quantity", "temperature", "time", "volume"] #https://github.com/facebook/duckling/blob/master/Duckling/Dimensions/FR.hs
locale: "fr_FR"
timezone: "Europe/Paris"
timeout: 0.2
# # No configuration for the NLU pipeline was provided. The following default pipeline was used to train your model.
# # If you'd like to customize it, uncomment and adjust the pipeline.
# # See https://rasa.com/docs/rasa/tuning-your-model for more information.
# - name: WhitespaceTokenizer
# - name: RegexFeaturizer
# - name: LexicalSyntacticFeaturizer
# - name: CountVectorsFeaturizer
# - name: CountVectorsFeaturizer
# analyzer: char_wb
# min_ngram: 1
# max_ngram: 4
# - name: DIETClassifier
# epochs: 100
# - name: EntitySynonymMapper
# - name: ResponseSelector
# epochs: 100
# - name: FallbackClassifier
# threshold: 0.3
# ambiguity_threshold: 0.1
# Configuration for Rasa Core.
# https://rasa.com/docs/rasa/core/policies/
policies:
# No configuration for policies was provided. The following default policies were used to train your model.
# If you'd like to customize them, uncomment and adjust the policies.
# See https://rasa.com/docs/rasa/policies for more information.
- name: MemoizationPolicy #AugmentedMemoizationPolicy
max_history: 0
priority: 6
- name: TEDPolicy
max_history: 0
epochs: 200
split_entities_by_comma:
special_type: False
any_information: False
constrain_similarities: true
model_confidence: softmax
entity_recognition: False
- name: RulePolicy
priority: 3
nlu_threshold_fallback: 0.3
core_threshold_fallback: 0.3
ambiguity_threshold: 0.01
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: True
nlu.yml (447.1 KB) rules.yml (726 Bytes) stories.yml (5.9 KB) cities.yml (88.1 KB) domain.yml (27.0 KB) actions.py (1.7 KB)
So I am sorry, my files are in french but it doesn’t matter if you don’t understand French.
My problem is when I said to my Chatbot :
Question : Quel est le planning
Answer : utter_activities
Question : Quel est le programme des activitiés
Answer : utter_activities
But if I said :
Question : Quel est le programme
Answer :
Why my synonym “planning” doesn’t match in the last case ? And why I got and not the utter_defaut answer ?
- synonym: planning
examples: |
- planning des activités
- programme
- programme des activités
Another thing I don’t understand is for the word before and after an entity. For example, if I said :
Question : Quelle est la météo à Chambéry
Answer : utter_weather with city = eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkNoYW1iw6lyeSJ9fQ==
Question : Quelle est la météo de Chambéry
Answer : utter_weather with city = eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkNoYW1iw6lyeSJ9fQ==
But if I said :
Question : Quelle est la météo à Annecy
Answer : utter_weather with city = eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkFubmVjeSJ9fQ==
Question : Quelle est la météo de Annecy
Answer : utter_weather with city = None
Question : Quelle est la météo de Annecy stp
Answer : utter_weather with city = eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkFubmVjeSJ9fQ==
It’s very strange ?
Nota : the key are encoded in base 64 ({“group”:“city”,“params”:{“city”:“Annecy”}} for exemple)
- synonym: stp
examples: |
- s'il te plaît
Same thing, for my synonym “stp”, why when I said :
Question : Quelle est la météo à Chambéry stp
Answer : utter_weather with city = eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkNoYW1iw6lyeSJ9fQ==
But if I said :
Question : Quelle est la météo à Chambéry s’il te plaît
Answer : utter_default
This is the output of my console for this example :
2021-09-10 17:59:09 DEBUG rasa.core.lock_store - Issuing ticket for conversation 'test_user'.
2021-09-10 17:59:09 DEBUG rasa.core.lock_store - Acquiring lock for conversation 'test_user'.
2021-09-10 17:59:09 DEBUG rasa.core.lock_store - Acquired lock for conversation 'test_user'.
2021-09-10 17:59:09 DEBUG rasa.core.tracker_store - Recreating tracker for id 'test_user'
2021-09-10 17:59:09 DEBUG rasa.nlu.selectors.response_selector - Adding following selector key to message property: chitchat
2021-09-10 17:59:09 DEBUG urllib3.connectionpool - Starting new HTTP connection (1): 172.17.0.2:8000
2021-09-10 17:59:09 DEBUG urllib3.connectionpool - http://172.17.0.2:8000 "POST /parse HTTP/1.1" 200 None
2021-09-10 17:59:09 DEBUG rasa.core.processor - Received user message 'quelle est la météo à Chambéry s'il te plaît' with intent '{'name': 'say_special_type', 'confidence': 0.9999999602720443}' and entities '[{'entity': 'city', 'start': 22, 'end': 30, 'confidence_entity': 0.8812001804963028, 'value': 'eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkNoYW1iw6lyeSJ9fQ==', 'extractor': 'CRFEntityExtractor', 'processors': ['EntitySynonymMapper']}, {'entity': 'special_type', 'start': 11, 'end': 19, 'confidence_entity': 0.9991491902140056, 'value': 'type_weather', 'extractor': 'CRFEntityExtractor', 'processors': ['EntitySynonymMapper']}, {'entity': 'special_type', 'start': 34, 'end': 42, 'confidence_entity': 0.5589575690818338, 'value': 'te plaît', 'extractor': 'CRFEntityExtractor'}]'
2021-09-10 17:59:09 DEBUG rasa.core.processor - Current slot values:
PER: None
city: eyJncm91cCI6ImNpdHkiLCJwYXJhbXMiOnsiY2l0eSI6IkNoYW1iw6lyeSJ9fQ==
time: None
any_information: None
special_type: te plaît
session_started_metadata: None
2021-09-10 17:59:09 DEBUG rasa.core.processor - Logged UserUtterance - tracker now has 152 events.
2021-09-10 17:59:09 DEBUG rasa.core.policies.memoization - Current tracker state:
[state 0] slots: {'any_information': (1.0,), 'special_type': (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0)}
[state 1] user intent: say_special_type | user entities: ('city', 'special_type') | previous action name: action_listen | slots: {'any_information': (1.0,), 'special_type': (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0)}
2021-09-10 17:59:09 DEBUG rasa.core.policies.memoization - There is no memorised next action
2021-09-10 17:59:09 DEBUG rasa.core.policies.ted_policy - TED predicted 'utter_weather' based on user intent.
2021-09-10 17:59:09 DEBUG rasa.core.policies.rule_policy - Current tracker state:
[state 0] slots: {'any_information': (1.0,), 'special_type': (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0)}
[state 1] user text: quelle est la météo à Chambéry s'il te plaît | previous action name: action_listen | slots: {'any_information': (1.0,), 'special_type': (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0)}
2021-09-10 17:59:09 DEBUG rasa.core.policies.rule_policy - There is no applicable rule.
2021-09-10 17:59:09 DEBUG rasa.core.policies.rule_policy - Current tracker state:
[state 0] slots: {'any_information': (1.0,), 'special_type': (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0)}
[state 1] user intent: say_special_type | user entities: ('city', 'special_type') | previous action name: action_listen | slots: {'any_information': (1.0,), 'special_type': (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 1.0)}
2021-09-10 17:59:09 DEBUG rasa.core.policies.rule_policy - There is no applicable rule.
2021-09-10 17:59:09 DEBUG rasa.core.policies.ensemble - Made prediction using user intent.
2021-09-10 17:59:09 DEBUG rasa.core.policies.ensemble - Added `DefinePrevUserUtteredFeaturization(False)` event.
2021-09-10 17:59:09 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_2_RulePolicy.
2021-09-10 17:59:09 DEBUG rasa.core.processor - Predicted next action 'action_default_fallback' with confidence 0.30.
2021-09-10 17:59:09 DEBUG rasa.core.processor - Policy prediction ended with events '[<rasa.shared.core.events.DefinePrevUserUtteredFeaturization object at 0x7f7e0c31da20>]'.
2021-09-10 17:59:09 DEBUG rasa.core.processor - Action 'action_default_fallback' ended with events '[BotUttered('None', {"elements": null, "quick_replies": null, "buttons": null, "attachment": null, "image": null, "custom": {"blocks": [{"type": {"name": "not_understood", "parameters": {"pepperSpeech": [{"pepperText": {"say": "Je n'arrive pas \u00e0 comprendre ce que vous voulez faire. Pourriez-vous reformuler ?"}}, {"pepperText": {"say": "Je suis d\u00e9sol\u00e9 mais je ne comprends pas votre demande. Pourriez-vous reformuler ?"}}, {"pepperText": {"say": "D\u00e9sol\u00e9 je ne peux pas r\u00e9pondre \u00e0 cette demande."}}]}}}]}}, {"utter_action": "utter_default"}, 1631289549.4567022), <rasa.shared.core.events.UserUtteranceReverted object at 0x7f7dcc45acc0>]'.
2021-09-10 17:59:09 DEBUG rasa.core.processor - Current slot values:
PER: None
city: None
time: None
any_information: None
special_type: __other__
session_started_metadata: None
2021-09-10 17:59:09 DEBUG rasa.core.processor - Predicted next action 'action_listen' with confidence 1.00.
2021-09-10 17:59:09 DEBUG rasa.core.processor - Policy prediction ended with events '[]'.
2021-09-10 17:59:09 DEBUG rasa.core.processor - Action 'action_listen' ended with events '[]'.
2021-09-10 17:59:09 DEBUG rasa.core.lock_store - Deleted lock for conversation 'test_user'.
And one more thing,I have tested with DietClassifier but the result isn’t better. And my command to train and run is :
rasa train --augmentation 0 --force --debug && rasa run --debug -p 5004
Thank you for your help.
Arnaud
But, Let me try help you, but for that you need to see some videos and GitHub repo for the code, I hope that is fine with you?