I need help in nlu data extraction(ner_cref)

Rasa Open Source
1

I have two type of entity(return date and departure date) but rasa nlu can’t figure which is departure date and which is return date eg:- “can you help me to find flight from goa to chennai on 10 december 2019” this i my sentence here 10 december is departure entity in training set but after train it give return date Please help me

2

@dharmu Can you please share your NLU data? Also what Rasa version are you using?

3

{ “id”: “a499b5b4-ccfe-4cc2-96ba-75b258958527”, “data”: [ { “text”: "can i book flight from ", “userDefined”: false }, { “text”: “Delhi”, “alias”: “from”, “meta”: “@sys.location”, “userDefined”: false }, { “text”: " to ", “userDefined”: false }, { “text”: “Chennai”, “alias”: “to”, “meta”: “@sys.location”, “userDefined”: false }, { “text”: " on ", “userDefined”: false }, { “text”: “23rd july”, “alias”: “departure”, “meta”: “@sys.date-time”, “userDefined”: true }, { “text”: ", ", “userDefined”: false }, { “text”: “return”, “meta”: “@sys.ignore”, “userDefined”: false }, { “text”: " on ", “userDefined”: false }, { “text”: “June 12”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “cd3de7f1-d811-4ccb-a603-372f64e988f9”, “data”: [ { “text”: “returning”, “meta”: “@sys.ignore”, “userDefined”: false }, { “text”: " on ", “userDefined”: false }, { “text”: “November 10”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “0258ef7d-0e1c-4570-9922-98952792afec”, “data”: [ { “text”: “return”, “meta”: “@sys.ignore”, “userDefined”: false }, { “text”: " on ", “userDefined”: false }, { “text”: “23rd June”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “6b745f8c-3910-4491-9ffe-951c92c9c740”, “data”: [ { “text”: "I will return on ", “userDefined”: false }, { “text”: “Monday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “9014a390-5f24-466d-8ad6-a2436d60a8fd”, “data”: [ { “text”: "I fly back on ", “userDefined”: false }, { “text”: “Monday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “f66bdd5e-f8cd-413a-b8cd-c1d13afcec34”, “data”: [ { “text”: "the date I return is ", “userDefined”: false }, { “text”: “Tuesday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “2a6e9d2c-7c91-4b2b-9c74-fa615c941ea1”, “data”: [ { “text”: "I will return ", “userDefined”: false }, { “text”: “tomorrow”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “9b3e244c-f819-46df-a839-b44ba772b441”, “data”: [ { “text”: "I\u0027m back on ", “userDefined”: false }, { “text”: “Friday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “21b30683-db49-436f-9746-b8157564e7ae”, “data”: [ { “text”: "ok the return is ", “userDefined”: false }, { “text”: “next Monday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “56b958d7-ae47-49bc-8cf1-c0ded3a45679”, “data”: [ { “text”: "getting back on ", “userDefined”: false }, { “text”: “Tuesday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “7a90def5-4fc2-484e-b313-ee976b36a8cf”, “data”: [ { “text”: "I\u0027m going to fly back on ", “userDefined”: false }, { “text”: “Monday”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: true } ], “isTemplate”: false, “count”: 0 }, { “id”: “08bc1eb0-9ecc-4230-a3a7-6ef88fa8ff27”, “data”: [ { “text”: "actually ", “userDefined”: false }, { “text”: “returning”, “meta”: “@sys.ignore”, “userDefined”: false }, { “text”: " ", “userDefined”: false }, { “text”: “July 25”, “alias”: “return”, “meta”: “@sys.date-time”, “userDefined”: false } ], “isTemplate”: false, “count”: 0 },

4

something look like that and i used rasa_nlu == 0.11.4

5

You are using quite an old version. Are you considering updating to Rasa 1.0? We improved quite a lot over time.

Your training data schema looks strange. Did you checked our documentation on how the training data should be formatted? (Training Data Format — rasa NLU 0.11.4 documentation)

What pipeline are you using?

In general, you should make sure that you have enough examples in your training data for both kind of dates. How much training data do you have per date type?

6

i used dialogflow data and i have enough data for training but i try on new version then it give some result no any improvement in nlu model.