janaka1984
(Janaka1984)
October 22, 2019, 3:56pm
1
when integrating rasa tracker with MongoDB, why chatbot is not correctly classified intent. it returns wrong utter
This is my configuration
language: "en"
pipeline:
- name: "tokenizer_whitespace"
- name: "ner_crf"
- name: "ner_synonyms"
- name: "intent_featurizer_count_vectors"
- name: "intent_classifier_tensorflow_embedding"
policies:
- nlu_threshold: 0.3
core_threshold: 0.3
fallback_core_action_name: "action_default_fallback"
fallback_nlu_action_name: "action_default_fallback"
deny_suggestion_intent_name: "out_of_scope"
imLew
(Nikolai)
October 24, 2019, 10:05am
2
Hi janaka,
the tracker/database should have no effect on the classification. Try retraining your bot and testing in different situations.
janaka1984
(Janaka1984)
October 24, 2019, 10:33am
3
Continuing the discussion from Chat bot is not correctly classify intent :
Hi @imLew
then how tracker and mongoDB effect chatbot?
is it affect to predict response by referring history of DB data
imLew
(Nikolai)
October 25, 2019, 9:19am
4
The bot responses are indeed based on the events in the history, which is stored in the tracker. However which tracker store you choose should not affect the behavior of the bot.