Actions for User Messages

You can edit, flag, or deactivate the user messages added in the User Messages section of the step.

• To edit the user messages, you can directly start to make changes.
• When you flag a user message, that particular user message appears as an autocomplete suggestion to the user. Read here to know more about auto-suggestions.
• You can deactivate a user message when you don't want the bot to get trained on an already added user message.

To understand the user's query bot supports spelling mistakes and tries to capture the correct intent using entities. There are two modules that help the bot to achieve this -

• Spell Correction
• Entity Tagging

Spell Correction

To account for the user’s mistakes, the Sentence Similarity module accounts for some spelling mistakes that a user could make. For example, if a user types “Show me services” or “Show me services”, they would still be matched to the “Show Services” step as per the above examples.

• Spell checker

  User messages should not contain common spelling mistakes. The spell corrector module will take care of these scenarios.

  Wrong	Right
  thanx | thankss | thankks	thanks
  rechrg | rechrge | recharg	recharge
  bouns | bonas	bonus
  case bones | cash bones	cash bonus

Spell checker consists of a normalization module which deals with scenarios such as -

Entity Tagging

With the feature entity tagging in user messages, bot builders no longer have to add similar variations in the user message with all the entity values mentioned in the entity dictionary.

Instead, they have to enter one user message variation with the entity value. The entities get tagged in the user message if they are added to the dictionary.

How does this work?

Entities are auto-tagged in the user message if the entity values are populated in the entity dictionary.

As long as one entity value is tagged in the user messages, the system will automatically define user message variations with other entity values.

For example, You might have steps that have sentences like "I want to play bahubali on Jio cinema", “I want to play bahubali on Hotstar” etc. on the ‘Play_Content’ step.

If you notice these utterances, the only difference is "Jio cinema" and “Hotstar” which can be added as values in the same entity.

To simplify and reduce the effort of generating such User Responses where a common set of words that belong to the same entity are needed, you can enable the entity feature on the bot.