FallbackExample

Failing Gracefuuly

This is the example bot to demonstrate the fallback scenarios provided by Rasa. Its sample code from the post on Fallback Strategies, which can be found on Medium.com.

It builds upon our existing example, which can be found here.

Improvements over Chatbot v2

• addition of fallback mechanisms
• addition of simple, single stage and two stage fallback mechanisms
• support for asking the user to rephrase themselves
• support for asking the user to select from some suggestions

A few examples:

Simple Fallback

👨 : !@#?*& (something the bot doesn't understand)

🤖 : Sorry! Let me connect you to a human...

Single-Stage Fallback

👨 : !@#?*& (something the bot doesn't understand)

🤖 : Sorry! What do you want to do?
    (a) Supply Contact information
    (b) Agree
    (c) Disagree
    (d) End the converation
    (e) None of these

👨 : (e) None of these (button click)

🤖 : Sorry! Let me connect you to a human...

Two-stage fallback

👨 : !@#?*& (something the bot doesn't understand)

🤖 : Sorry! What do you want to do?
    (a) Supply Contact information
    (b) Agree
    (c) Disagree
    (d) End the converation
    (e) None of these

👨 : (e) None of these (button click)

🤖 : I'm sorry, I didn't quite understand that. Could you rephrase?

👨 : !@#?*& (again, something the bot doesn't understand)

🤖 : Sorry! What do you want to do?
    (a) Supply Contact information
    (b) Agree
    (c) Disagree
    (d) End the converation
    (e) None of these

👨 : (e) None of these (button click)

🤖 : Sorry! Let me connect you to a human...

Contents

This project follows the format of a standard Rasa project. There's a directory called data for training data like nlu, stories, and rules.

There's a directory called actions, which contains all your custom actions.

You'll also find the domain.yml file, which mentions all your intents, entities, slots, responses and actions.

Finally, there's the config.yml file, which specifies the components your bot is comprised of.

The config file has a modification compared to the previous versions of this bot: low training epochs for the DIETClassifier. This is to simulate fallback behaviour. If you want to build a properly functioning bot, increase epochs to about 100.

Usage

1. Clone this repo
2. Navigate to the RasaChatbot directory
3. Install rasa>=2.6.2 in an env.

Modify the files in data/ or the domain.yml file to play around.

Training the bot

Validating the data

Before training the bot, a good practice is to check for any inconsistencies in the stories and rules, though in a project this simple, it's unlikely to occur.

$ rasa data validate

Training

To train the bot, we simply use the rasa train command. We'll provide a name to the model for better organization, but it's not necessary.

$ rasa train --fixed-model-name contact_bot_v2

Chatting with the bot

To test your bot, open a new terminal window and start a rasa shell session.

$ rasa shell

This will let you chat with your bot in your terminal. If you want a more interactive UI and a little more debugging information like what intents were identified and what entities were extracted, you can use Rasa X.

Testing the various fallback strategies

To test the different fallback strategies, simply comment out the rules for all except for the one you want to test. These rules are located along with the regular rules in rules.yml.

After this, retrain the bot to test the strategy. Be sure to keep the epochs for your DIETClassifier low (below 5), to simulate nlu_fallback reliably.

---

You can find me on medium here.