When some dates are given in the utterance in the format mm/dd e,g,. 9/11 is not picking up by the date entity , whereas 2/11 is getting picked up.
below are the utterances
“reschedule job 123456 to tech 1231 to 2/11 at 3pm” – picking up
“reschedule job 123456 to tech 1231 to 9/11 at 3pm” – not picking up
In Debug Log → NLP analysis for utterance “reschedule job 123456 to tech 1231 to 9/11 at 3pm” the date part is getting divided i.e., 9/11 = 0.81818181818181823228286475568893365561962127685547.
what to do ??
Based on my experience building bots with complex entity extraction, I would discourage you from using NLP in the scenario you have described. This is a PERFECT use case for a GenAI (Agent) node and I would suggest you redesign your dialog flow to use this capability. If you cannot use this node type, I’ll give you two suggestions that might help:
- Make sure the entity that you’re using to collect “9/11” is a date type and not a number or string.
- Consider using a Composite entity rather than separate entities. This will allow you to define entity patterns and may improve accuracy.
FYI… I was unable to get consistent entity capture from an utterance that is so complex. There are just too many variations in the way a user can format their entry (9/11, 9-11, 9-11-24, “next Tuesday”). However, I had great success using the GenAI node in similar scenarios.
Good luck!
