Improving skills with Attribute Tagging

Owned by Bhavani.tvr (Unlicensed)
Last updated: Feb 05, 2021 by sneha.tripathi
2 min read

Attribute Tagging in Luma Virtual Agent is used to make your bot intelligent enough to understand the user phrases, identify the information and assign the possible values to the next prompts in the skill. It enables the Bot to extract and assign the values to the attributes on its own and skip the prompts if the values are already provided by the end-user. This can be achieved by tagging a word in a user phrase and linking it to an attribute.

There are two ways to tag an attribute:

  1. NLP extraction 

  2. Entity extraction  

NLP extraction  

NLP extraction is available only for attributes of the TEXT data type. In this method, the bot can extract Text type attributes from the user's phrase when their NLP Extraction Method is set to Value or Pattern. 

Follow the below steps to enable NLP extraction for an attribute.

Enable NLP Extraction in Attribute definition:

  1. Navigate to Bot Menu → Skill builder → Attributes.

  2. If enabling NLP extraction for an existing attribute, select and view attribute details. Click Edit.

  3. For a new attribute, Click on Create Attribute and attribute details as required.

  4. Set NLP Extraction Method to Value or Pattern Matching as required.

  5. Add Value and Synonyms as required. This enables the bot to understand the user's input as a synonym and aligns it to the value.

  6. Click on Save.

Tag value to an attribute :

  1. Navigate to Bot Menu → Skill builder → Skills.

  2. Select the required skill or create a new skill.

  3. In the Invocation Type section, add the required user phrase.

  4. Double-click on the word to be tagged or highlight a group of consecutive words in a user phrase that you want to tag to an attribute. Select the required attribute from the dropdown.
    The word(s) appear as the Tagged Value.

  5. If Value and Synonyms are configured, select the Resolved value that should be assigned to the actual Attribute.

  6. Complete the skill configuration and Click Save.

To train the Bot to identify and extract information from the given user’s input phrase, add as many phrases as you can. It is advised to add 20-25 phrases with various permutations and attribute tagging. This enables the bot to identify patterns and extract information from unseen phrases as well.

Entity Extraction 

Entity Extraction occurs automatically for attributes with datatype EMAIL, NUMBER, and TIME. The bot automatically extracts the values from the user's phrase and assigns the value to the attributes according to the data type in the sequence defined in the Conversation flow. 

This does not require any special configuration.