Object parsing with rasa for chatbot applications

Rasa Open Source
1

Hello, So I came up with a problem in which I am trying to create a chatbot for various product information automatization.

So what I am trying to do: For example I have 5 products. (Mixer, Refrigerator, Toaster, Microwave and Electric Grill) All these products have in some way same information:

For ex:

{
    "Microwave":{power:500w, size:50x30x50, plug:EU},
    "Refrigerator":{power:2000w, size:100x50x100, plug US},
    "Toaster":{power:1000w,size:20x20x20,plug:EU}
    ...
}

And I want to make an chat assistant that could answer something simmilar to this situation:

  • User: Whats the appliance power?
  • Bot: Okay, whats the applience you have bought from us?
  • User: An Microwave
  • Bot: A our XNZ Microwave uses 500watts of power
  • User: Thank you…

So Im in a pickle because I dont have any idea what would be the smartest way to have this “product” information saved and trained for rasa. I guess I could always write a ton of utter responses and intents for each of the applience questions. But I am trying to think of a smarter way to for example have only one story (“Whats the power?”) and let rasa handle the rest.

Thank you!

2

You can do this in two ways. One is through Knowledge bases and another one is to build a contexual bot through custom actions + slots + stories.

Thanks Uday

3

Any sources or tutorials that I could follow up?

4

For Knowledge base, you can follow the below repo.

Whereas for Contexual bot, it entirely depends on the stories and custom actions by using slots at appropriate places. So, there is no use case for your example. But it can be done.

Follow this blog for second use case.

5

I found this one. In Rasa documentation

6

Both are examples of Knowledge based bots. You can use anyone of this.

In the link that I have provided, there are more clear instructions on how to build a knowledge based bot if you observe.