How to use location to get the right intent

For example, court a:intent: gotomoviea watches movies; The folder of court B also has intent: gotomovieb watches movies; The folder of court C also has intent: gotomoviec watching movies; Now I’m in court C, saying I watch movies; Identification is intent: gotomoviec. How to force the identification intention is in the C court; It may be recognize as another court now.In my Rasa,now I need change the recognize intent to the court’s similar intent. I want to use the location of different courts to force it recognize the intent to the current court’s intent.


There are 600 courts, but each courts have their own data or corpus. and some intent in one court is similar or same with other courts. the questions or words in examples are same.but the intent_name is not different. It takes too long time to gather the similar intent_name to same column,because I need to compare each other similar intents.

I want the different courts use their own data or corpus.and when say the similar intent, The identified intention is in the current court, not other courts.

I can only change the identified intention after the rasa recognize it. but now it not from the beginning it can recognze to the intent of current court.How to use the slot : “location” to do it. I use flask interface to get the location.


how to use the extra variable “location” to in the Rasa,and In my Rasa, it contain 600 courts, each court in each folder.There maybe some intent in nlu.yml of different courts is same or similar, how can recognize the intent is current court. if I in court “lawbot_shushan” ,location is :“shushan”. when the intent is similar, how to recognize to the intent of folder “lawbot_shushan” , not other folders. now it may recognize to one court not the court in “shushan”.I need th change the intent to similar intent in court “shushan” folder “lawbot_shushan”.

so how recognzie the intent from the beginning is the intent of court “shushan”. the current court. not other similar intent of other courts

now I can ony train one model. not for each court train each model. Or if train each model of 600 courts in one computer not slow and take time not more than 10 hours, it alse can use this way. but I think it difficult to train fast in one computer for 600 courts ,so I use one model


how to use the {“location”:“Beijing”,“msg”:“hello”}
how to include the location to the recognize process?

where to change the recognize process

how to introduce and use reference variables to the recognize process? Thanks


or is there like the “current_state = tracker.current_state()” that can get a lots of information about the similar intent, but here I only get the similar intent, which way can get the similar intent and the intent form which file or folder.

This is the current_state

'lawq': {
			'response': {
				'responses': [{
					'text': 'It belongs to the exclusive jurisdiction stipulated by law and is under the jurisdiction of the court where the port is located'
				'confidence': 0.07165741175413132,
				'intent_response_key': 'lawq/ask_portoperationcase_determinejurisdiction',
				'utter_action': 'utter_lawq/ask_portoperationcase_determinejurisdiction'
			'ranking': [{
				'confidence': 0.07165741175413132,
				'intent_response_key': 'lawq/ask_portoperationcase_determinejurisdiction'
			}, {
				'confidence': 0.061188049614429474,
				'intent_response_key': 'lawq/ask_courtjurisdiction_objection_presentation'
			}, {
				'confidence': 0.05191769450902939,
				'intent_response_key': 'lawq/ask_jurisdictionalobjection_court_handle'
			}, {
				'confidence': 0.05186884477734566,
				'intent_response_key': 'lawq/ask_courtjurisdiction_objection'
			}, {
				'confidence': 0.04837268218398094,
				'intent_response_key': 'lawq/ask_party_jurisdiction_objection'
			}, {
				'confidence': 0.03890473395586014,
				'intent_response_key': 'lawq/ask_intermediatecourt_jurisdictioncase'
			}, {
				'confidence': 0.03637944534420967,
				'intent_response_key': 'lawq/ask_execute_jurisdictionalprotest'
			}, {
				'confidence': 0.03458021953701973,
				'intent_response_key': 'lawq/ask_infringementdisputecase_determinationofjurisdiction'
			'ranking': [{
				'confidence': 0.12068066745996475,
				'intent_response_key': 'lawq/ask_ontheplaintiff_situationofthedefendant'
			}, {
				'confidence': 0.1150425598025322,
				'intent_response_key': 'lawq/ask_withregardtotheplaintiff_situationofthedefendant'
			}, {
				'confidence': 0.0563642755150795,
				'intent_response_key': 'lawq/ask_issuance_judgment_time'
			}, {
				'confidence': 0.056098874658346176,
				'intent_response_key': 'lawq/ask_appeal_range'
			}, {
				'confidence': 0.048957616090774536,
				'intent_response_key': 'lawq/ask_withdrawalofapplication'
			}, {
				'confidence': 0.036208756268024445,
				'intent_response_key': 'lawq/ask_repeated_prosecution_definition'
			}, {
				'confidence': 0.03417505323886871,
				'intent_response_key': 'lawq/ask_legalrepresentative_party_difference'
			}, {
				'confidence': 0.03405805304646492,
				'intent_response_key': 'lawq/ask_missingperson_propertytrustee_change'
			}, {
				'confidence': 0.03375423327088356,
				'intent_response_key': 'lawq/ask_appeal_term'
			}, {
				'confidence': 0.029057178646326065,
				'intent_response_key': 'lawq/ask_litigation_capacity_party_handle'

but it have the 10 simialr intents , but In my rasa,it may 100,200,300 and so on. how can list all the similar intents.

it have the 10 simialr intents , how can list all the similar intents.

  • name: DIETClassifier epochs: 100 batch_size: 8 ranking_length: 100

ranking_length: 600


how to include the location to the recognize process?. .

I dont’t know, maybe now it not deveolp still.So I only can change the intent to the simiar intent in local court