Slot Reset Causing issues with next action prediction in rasa chat bot

Rasa Open Source
1

Please help me on the issue given below. Where Slot resetting is causing some weird problem.

My Slot Reset action look like below > class ActionRenew(Action):

    def name(self):
            return "action_renew"

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

My Stories looks like below

  • scarfFAQ{“issue”:“”, “key”:“”}
- utter_wait
- action_query_mongo
- action_renew

Rasa Logs are given below

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.processor - Received user message ‘what is SCARF?’ with intent ‘{‘name’: ‘scarfFAQ’, ‘confidence’: 0.9831768274307251}’ and entities ‘[{‘start’: 8, ‘end’: 13, ‘value’: ‘SCARF’, ‘entity’: ‘issue’, ‘confidence’: 0.9412032215449633, ‘extractor’: ‘CRFEntityExtractor’, ‘processors’: [‘EntitySynonymMapper’]}]’

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.processor - Current slot values:

core-rasa | disambiguation_message: None

core-rasa | issue: SCARF

core-rasa | key: None

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.processor - Logged UserUtterance - tracker now has 10 events

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘783a65a2-cf03-49fc-b659-43accf7054b3’.

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.policies.fallback - NLU confidence threshold met, confidence of fallback action set to core threshold (0.3).

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.policies.memoization - Current tracker state [{‘prev_action_listen’: 1.0}, {‘prev_utter_greet’: 1.0}, {‘intent_chitchat.greet’: 1.0, ‘prev_action_listen’: 1.0}, {‘prev_utter_greet’: 1.0, ‘intent_chitchat.greet’: 1.0}, {‘prev_action_listen’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}]

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.policies.memoization - There is no memorised next action

core-rasa | 2019-11-27 06:11:18 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_2_KerasPolicy

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.nlg.callback - Requesting NLG for utter_wait from http://x.x.x.x:8081/project/bf/nlg.

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.processor - Action ‘utter_wait’ ended with events ‘[‘BotUttered(text: One moment please…, data: {“elements”: null, “quick_replies”: null, “buttons”: null, “attachment”: null, “image”: null, “custom”: null}, metadata: {})’]’

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘783a65a2-cf03-49fc-b659-43accf7054b3’.

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.policies.fallback - NLU confidence threshold met, confidence of fallback action set to core threshold (0.3).

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.policies.memoization - Current tracker state [{‘prev_utter_greet’: 1.0}, {‘intent_chitchat.greet’: 1.0, ‘prev_action_listen’: 1.0}, {‘prev_utter_greet’: 1.0, ‘intent_chitchat.greet’: 1.0}, {‘prev_action_listen’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}, {‘prev_utter_wait’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}]

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.policies.memoization - There is no memorised next action

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_2_KerasPolicy

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.actions.action - Calling action endpoint to run action ‘action_query_mongo’.

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.processor - Action ‘action_query_mongo’ ended with events ‘[‘BotUttered(text: $MESSAGE$:SCARF (Application Reference Framework) drives InnerSourcing in the organization, data: {“elements”: null, “quick_replies”: null, “buttons”: null, “attachment”: null, “image”: null, “custom”: null}, metadata: {})’]’

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.processor - Current slot values:

core-rasa | disambiguation_message: None

core-rasa | issue: SCARF

core-rasa | key: None

core-rasa | 2019-11-27 06:11:19 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘783a65a2-cf03-49fc-b659-43accf7054b3’.

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.policies.fallback - NLU confidence threshold met, confidence of fallback action set to core threshold (0.3).

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.policies.memoization - Current tracker state [{‘intent_chitchat.greet’: 1.0, ‘prev_action_listen’: 1.0}, {‘prev_utter_greet’: 1.0, ‘intent_chitchat.greet’: 1.0}, {‘prev_action_listen’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}, {‘prev_utter_wait’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}, {‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘prev_action_query_mongo’: 1.0, ‘entity_issue’: 1.0}]

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.policies.memoization - There is no memorised next action

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_2_KerasPolicy

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.actions.action - Calling action endpoint to run action ‘action_renew’.

Here Action renew was called successfully.

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.processor - Action ‘action_renew’ ended with events ‘[‘AllSlotsReset()’]’

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.processor - Current slot values: core-rasa | disambiguation_message: None core-rasa | issue: None core-rasa | key: None

Slot reset happened successfully

core-rasa | 2019-11-27 06:11:20 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘783a65a2-cf03-49fc-b659-43accf7054b3’.

**Heart of the problem is given below. Rasa got the new message and interpreted it correctly with the right intent (scarfFAQ) and entities (issue : SCARF, key : None). But it did not follow through with the right action prediction which in this case should have been action_query_mongo .

Would really help if somebody can help me understand this strange behaviour. Thanks in advance**

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.processor - Received user message ‘what is SCARF?’ with intent ‘{‘name’: ‘scarfFAQ’, ‘confidence’: 0.9831768274307251}’ and entities ‘[{‘start’: 8, ‘end’: 13, ‘value’: ‘SCARF’, ‘entity’: ‘issue’, ‘confidence’: 0.9412032215449633, ‘extractor’: ‘CRFEntityExtractor’, ‘processors’: [‘EntitySynonymMapper’]}]’

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.processor - Current slot values:

core-rasa | disambiguation_message: None

core-rasa | issue: SCARF

core-rasa | key: None

Above Slots were picked up correctly. but finally I don’t get any answer

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.processor - Logged UserUtterance - tracker now has 18 events

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘783a65a2-cf03-49fc-b659-43accf7054b3’.

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.policies.fallback - NLU confidence threshold met, confidence of fallback action set to core threshold (0.3).

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.policies.memoization - Current tracker state [{‘prev_utter_greet’: 1.0, ‘intent_chitchat.greet’: 1.0}, {‘prev_action_listen’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}, {‘prev_utter_wait’: 1.0, ‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘entity_issue’: 1.0}, {‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘prev_action_query_mongo’: 1.0, ‘entity_issue’: 1.0}, {‘slot_issue_0’: 1.0, ‘intent_scarfFAQ’: 1.0, ‘prev_action_renew’: 1.0, ‘entity_issue’: 1.0}]

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.policies.memoization - There is no memorised next action

core-rasa | 2019-11-27 06:11:25 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_2_KerasPolicy

core-rasa | 2019-11-27 06:11:26 DEBUG rasa.core.processor - Action ‘action_listen’ ended with events ‘’

core-rasa | 2019-11-27 06:11:26 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘783a65a2-cf03-49fc-b659-43accf7054b3’.

My Policies are

policies:

  • name: FallbackPolicy nlu_threshold: 0.4 core_threshold: 0.3 fallback_action_name: “utter_unclear”
  • name: MemoizationPolicy
  • name: KerasPolicy featurizer:
    • name: MaxHistoryTrackerFeaturizer max_history: 5 state_featurizer:
      • name: BinarySingleStateFeaturizer

Any help in this issue would be really appreciated.

2

It looks like you don’t have a training story with action_renew in there, so core doesn’t know what to predict next

1 Like
3

Thanks Ghostvv for the help, but I am still not clear, because If you see the story above/below I have included action_renew in my story. Can you please explain more in details, that would be really helpful.

  • scarfFAQ{“issue”:“”, “key”:“”}
  • utter_wait
  • action_query_mongo
  • action_renew

Thanks again.

4

you need to add reset_slots event in your training stories, since that’s what your custom action returns