Not able to extract entity when its consists special character

[arindambose1010]

I am developing a bot, which need to handle special character also. But unfortunately it is not extracting that value. I’m providing the details of my config, nlu and regex… Could you please check and help me out to resolve this

config.yml -------------------- . recipe: default.v1 language: en pipeline:

• name: WhitespaceTokenizer intent_tokenization_flag: true case_sensitive: false intent_split_symbol: +
• name: RegexFeaturizer
• name: RegexEntityExtractor pattern: regex_patterns.yml # Define regex patterns for special characters
• name: CountVectorsFeaturizer
• name: DIETClassifier
• name: CRFEntityExtractor
• name: EntitySynonymMapper
• name: DucklingEntityExtractor
• name: CountVectorsFeaturizer analyzer: char_wb min_ngram: 1 max_ngram: 4
• name: FallbackClassifier threshold: 0.98 policies:
• name: MemoizationPolicy max_history: 50 epochs: 100
• name: TEDPolicy max_history: 50 epochs: 100
• name: RulePolicy
  assistant_id: 20230419-003217-calm-contract

nlu.yml

• intent: combine_case_cby examples: |
  • What’s the current status of case CPIB 10 from 2023 at Chandigarh?
  • Provide me with the details of CPIB 10 case for 2023 in Chandigarh.
  • Tell me the status of the 11 case under CPIB for 2023 at West Bengal.
  • How is 11 CP%28IB%29 case proceeding for 2023 at West Bengal?
  • 12 case in 2023 at Delhi related to CPIB.
  • Delhi 12 from 2023 associated with CPIB.
  • Delhi 12 2023 CP(IB) situation.
  • Chandigarh during 2023 regarding case 12.
  • What’s the current status of CPIB 11 case from 2023 at Chandigarh bench?
  • What are the details of CPIB 11 case in 2023 at Chandigarh bench?
  • Can you provide me the status of case CPIB 10 from 2023 at Chandigarh?
  • Tell me the status of CP(IB) 10 case for 2023 at Chandigarh.
  • How’s the status of CPIB 10 case going for 2023 at Chandigarh?
  • What’s the status of CPIB case number 10 of 2023 at Chandigarh?
  • Provide details about CPIB 10 case in 2023 at Chandigarh.
  • Tell me the status of 11 case under CP(IB) for 2023 at West Bengal.
  • 11 CP%28IB%29 case for 2023 at West Bengal.
  • 12 , 2023 Delhi for CPIB.
  • Delhi 12 , 2023 , CPIB.
  • Delhi 12 , 2023 , CP(IB).
  • Chandigarh of 2023 12.
  • status of CP%28IB%29 11 of 2023 of Chandigarh bench.
  • details of CP%28IB%29 11 of 2023 of Chandigarh bench.
  • what is the details of CPIB 10 of 2023 Chandigarh?
  • what is details of the CPIB 10 of 2023 Chandigarh?
  • what is the details of 2023 of CPIB 10 Chandigarh?
  • Status of CP%28IB%29 11 of 2023 of Chandigarh bench.
  • what is the status of CPIB 10 of 2023 Chandigarh?
  • what is status of the CPIB 10 of 2023 Chandigarh?
  • what is the status of 2023 of CPIB 10 Chandigarh?

regex_patterns.yml

version: “2.0”

extractions:

• name: “casetype” type: “regex” pattern: “[.*?]” # This pattern captures anything within square brackets

domain.yml

slots: casetype: type: text mappings: - type: from_entity entity: casetype entities: casetype

output of my console

Let me know if any thing more you required

[sonam]

you need to provide examples first, check out any entity extraction video on our YouTube.

[arindambose1010]

Hi Sonam, I’ve already used entity extraction in my several project. Now I am getting stuck where an entity consists special character. Entity always are getting None this time Please check the following images and take it as example. Please guide me to resolve it domain.yml

Config.yml

nlu.yml

[sonam]

Hi,

I think you need to change tokenizer, from whitespacetokenizer, or write a custom one. You can look into similar discussion here.

Rasa not picking special characters in an entity

As already pointed out by @ganeshv, we have a regex in place that splits words on those characters into separate tokens. So if you are using the WhitespaceTokenizer this will happen. If you want to keep the words, you can first of all try a different tokenizer or update the regex by writing a custom tokenizer (you can use the WhitespaceTokenizer as an example and just update the regex over there).