Regex Entity Extractor

Rasa Open Source
1

Hi I am trying to extract age using regex entity extractor and the issue that I am facing is that it is not being extracted. This is the way I have written it in my nlu file

  • regex: age_english examples: |
    • /^100|[1-9]?\d$/
  • intent: age_english examples: |
    • [25] (age_english)
    • [23] (age_english)

This is my story

  • story: Activate Form steps:
    • intent:-------------
    • action: -----------
    • action: user_details_form
    • slot_was_set:
      • loan_type:
      • name: ----------------
      • salary: ----------------
      • experience: ----------------
      • reason: ----------------
      • time: ----------------
      • age: 30

@ChrisRahme can you guide me on this?

2

Hello Afnaan,

Why is your regex so complicated for a number? Try \d{1,2}

Another culprit could be your formatting, you should not have a space between ] and ( when tagging entities. Try: [25](age_english)

3

hey @ChrisRahme , So my regex was complex I get that, I wanted age between 0 and 100. I put the space in between ) [ was because if I do not put it there so it is not displayed correctly on this site. After putting this in the nlu file do I have to make changes in the domain and stories as well? How do I identify what is the regex?

4

What are the changes?

If you just changed the Regex, you only have to retrain the bot.

5

I want to do this because my bot is getting confused between two slots

age and salary it overwrites the value of age in salary.

I think this is because both are numeric values.

salary: - entity: salary type: from_entity not_intent: age_english

age: - entity: age_english type: from_entity not_intent: salary_english I have written this as well. Can you please tell me some other way I might be able to do it? @ChrisRahme

6

Hmm yeah it’s confused because both are numbers.

Sorry, I don’t know how to solve this :sweat_smile: But this is most probably your problem. Please create a new thread with this problem clearly defined in the title. For example, “Extract numbers as different entities”, “Extract age and salary”, etc.

7

Linking the new thread in case anyone in the future needs it: