Want to filter any gibberish input or sentences that do not make any sense before putting through the chatbot, any ideas ??
@nik202 I am working on a project wherein I take the user input to my question and instead of giving options, i ask the user to type his/her response. Then I try to map the input to the backend options, in case the matching is not satisfactory (low confidence), i try to ask the person further questions on what category does his/her answer lies from, but this leads to bias in the person’s response as the category he/she enters might be different than his/her initial response as looking at the category he/she sees something that makes more sense. Thus due to this system some adversaries try to input in gibberish and try to get the categories, which needs to be avoided
What are your backend options? is it some APIs or Databases?
@Madhur-1 for this set of scenarios you can you the fallbacks either default fallback or two-stage fallback.