Use Case - Train the Chatbot

As part of you initial roll out of the Chatbot to your organization, you may want to include a "Train the Chatbot" campaign where users can submit their suggestions for use cases they would like to see enabled from the Chatbot.  This campaign can be implemented as a longer running campaign, unlike the Name the Chatbot campaign.

This use case leverages an Innovation Studio record and process definition as well as IBM Watson Assistant content. The record definition is use to store the training suggestions along with the users name that submitted the suggestion.  The process definition is called from the Dialog node, implemented within IBM Watson Assistant and is used to create the record.

BMC Helix Innovation Studio implementation details:

• Chatbot training suggestions record definition:
  When you install the Chatbot Starter Kit content package, you will find the Train the Bot Campaign record in Records tab.  This record definition has the standard record definition fields along with the following two fields:

• • User ID - used to store the users unique login ID
  • Bot Name - used to store the user's training suggestion 
    
    
• Chatbot training suggestions process definition:
  Within the Process tab you will see the Train the Bot Campaign process definition.  The process is designed to be called from IBM Watson Assistant in order to create a record in the Train the Bot Campaign record definition.

IBM Watson Assistant implementation details:

When you install the Chatbot Starter Kit content package the following IBM Watson Assistant content will be available. 

• Intents: 
  The #trainthebot intent comes with a number of utterances for this use case,  You can add / remove utterances as needed to support you implementation.
  
  
• Dialog nodes:
  The BMC-TrainTheBot parent dialog node comes with a predefined interaction model used by the Chatbot to ask and get information from the user about training the bot.  You can either keep the interaction text provided or you can make changes as needed.  You will however need to update the text as there is some OOB text given regarding the name the bot campaign rules, dates for the contest, etc.
  
  The following is an example of what the interaction looks like:
  
  
  
  
  Selecting Yes will continue with this request use case, while No will cancel this request.
  
  
  
  In this sample text you are laying out the contest rules as well as any additional information.   This text will need to be modified for your implementation.
  
  
  
  After giving a training suggestion, the user is prompted for a justification.  This can assist you and your team when trying to select a winner for the contest or it may provide some additional insights into the request.
  
  
  
  

After the user enters a justification the Chatbot submits the entry to the record definition previously mentioned above.  

You can run a report against the record definition to get a list training suggestions and the users that submitted them.