Detect Variables right on the Intent start

i´d like to create a task that replies utterance like:
User: I saw your product name X and would like to know more about it.
Bot: Great! X is a great product, bla bla bla

My main goal here is to receive names from the user inserted in some pattern of questions, and the bot replies a standart answer using the X name as a variable like the example above. Is it possible?

I thought about using variables or knowledge base. Or another method is better?

Hi @felipe,

You can use Named Entity Recognition approach in NLP Training.
Please refer the following link:
https://developer.kore.ai/docs/bots/concepts/natural-language-processing-nlp-guide/entity-detection/

Hope this helps.
Let us know if you need any further clarification on the above.

Regards,
Yoga Ramya.

Thank you @yogaramya.mendu

Using the same parttern
example:
I need support for Windows
I need support for MAC

How can I detect the different entities that I sell/support from those that I do not?

@felipe: Please check if the below configuration could help.

  1. Create an entity in the dialog (Support) with the entity type as list of item (enumerated)and have the allowed values configured in it. (Say Windows, MAC)
  2. For the utterance “I need support for Windows”, you could tag which Windows the entity value.
  3. Train the utterances.
  4. Pass the utterance “I need support for Windows” to the bot and the bot wouldn’t ask for entity value
  5. If the utterances passed is “I need support for Android”, the bot would ask for entity value as its not present in the values configured at entity node

Please refer to the below screenshots for reference. To explain the use case, we have configured a dialog (Breakfast) with entity list of values (bacon, cereals).

image

image

image

image

image

image

Thanks @Subrahmanyam

my final configuration would demand that:

  • the reply for windows is Reply X
  • the reply for android is Reply Y
    In short, I have to reply that I sell Windoes but do not sell Android.
    Can do it on the same task?

I tried to make different tasks for it but the bot is mixing the responses

Please check if the below configuration could help.

  1. Create the entity type as string to store the user input
  2. add a script node to compare user input against an array (Windows / MAC)
  3. In case user input is not present in the array, create a variable inside the context to set false at the script node
  4. Have transition condition from script node to present the message “I sell Windoes but do not sell Android”. using message node based on the context variable value (false)
  5. Now add a script node to delete the entity and link it again to the entity node.

thank you @Subrahmanyam

I was trying with a different aproach. What do you think of it bellow?

  1. Created the entity as Item List Enumerated and added all the items name with synonyms.
  2. Trained the Task to detect the name of the product on sentences like Do you sell Windows? or Do you have support for windows?
  3. Marked the above product names as Entity type on the training section
  4. On the task, I crearted a filter of if and els if directing the task to a different reply depending of the Entity is equal or not from the value defined on the item list.
  5. It was working fine with few produtcs and few utterances, but once I satarted to grow in complexity, it started to loose controle and mix answers.

I will try to do the method you presented above. Thank you