Universal Ai Diploma — IBM Cloud Assistant Recurrent Neural Network Alternative
Author of notes: God Bennett, Universal Ai Diploma.
This recurrent neural network based solution can take place in the browser without signup or visa card data unlike IBM cloud.
Rasa natural language works similar to ibm cloud, whereby we can configure both bot input method from user and bot response/.
I will discuss my modified Rasa collab (link above modified to remove setup errors and add medicine bot response/acceptance) in the instructions below:
1. Modify Section 1 (intents): These are topics/ideas of what the ai bot can be told by user, i.e. words the user types to chatbot.
Section 2. (intent/actions mapping/connection): This is what the bot can say back to the user.
i. Add intent from section 1
ii. Add new Chatbot Action you added in Section 2 — This is the bot’s responses to respective user words/intents typed to chatbot.
You can modify for example, by:
a. Changing intent (1) found in section 1 in string “ ## intent:intent_greet “, to intent:intent_sports_inquiry, and changing all examples from:
- hello there
- hello there
- good morning
- good evening
- how far
- hey there
- whats up
- hey dude
- good afternoon
to the following:
— Where can I get sports magazine?
— I need sports stuff
— I need sporting goods
— When can I get sporting goods?
— Where can I go to the latest football game?
— I need pills
The above tells rasa bot an idea of what user may say, and recurrent neural network should allow the bot to accept text
that is similar, and not necessarily the same, i.e. natural language input!!.
b. Changing the action related to the intent you changed, for eg, in section 2.b, “ utter_greet “, to “utter_sports_inquiry” and changing all examples from:
- text: “\n\n Universal Ai Diploma Bot: \n\n Hey! Welcome to NRC Kubwa terminal . How may I help you ? \n\n \n\n You:”
to the following:
- text: “\n\n Universal Ai Diploma Bot: \n\n Hey! Sporting goods can be found in your store close by, tomorrow. \n\n \n\n You:”
c. Changing the mapping of action/intent related to the event you changed:
2. Now that you’ve customized the potential bot action system by code in steps above, run all cells (press play button starting from first cell)
After you ran the first cell successfully you should see a tick and model saved message at bottom. (If not restart the shell by clicking a restart button as directed)
All cells must have a green tick as well as success messages at bottom of each cell, like the training cell below with my custom ..._medicine_inquiry data:
3. If you ran all cells successfully, you would have reached to the end where you can interact with your custom rasa bot, like seen in the image below where I modified the bot to include medicine demo response-ability!
Type a message related to your modified intents and utterances.
- Ensure your colab file is public by pressing share button and making available. Button is top right.
- Go to your github profile and create a new empty repository, calling it something like “Universal Ai Diploma Week 4 RNN Chatbot — Your service name”
- Save your modified colab file to your github repository by going to File / Save copy in github, and selecting your repository from (1) above: