4. Asking user info through input validation
You have now built the basic version of the Choo Choo bot that detects information in user messages, great! Now let's see how your bot can directly ask for user input with input validation dialogs
To complete our Choo Choo bot, we need to ask the following information from the user:
- 1.Origin
- 2.Destination
- 3.Departure date
- 4.Departure time
- 5.Class
This works great when the users says, for example:
- Tomorrow, I need to go from Amsterdam to Brussels at 2pm in second class
But right now, our bot can only detect this information when the user states it all in a single sentence. What then if the user doesn't provide those elements? Like in:
- I would like to book a train ticket
The last sentence doesn't leave the bot with all it needs to know. To overcome this issue, ChooChoo should be able to ask for for those pieces of information though Input Validation dialogs. An Input Validation is a dialog type where you can ask for a specific thing and save the answer directly into a variable to be reused later.
By using Input Validations, the bot can ask for specific pieces information and stores them as variables.
Chatlayer offers 4 dialog types. Since the start of this tutorial, we have only been using Bot Messages. The three other ones are Input Validations, Go-To's and Action Dialogs. Learn more about them here.
Let's dig in and add some Input Validations to ChooChoo 👇
We are going to create new Input Validations. This can be done in the main flow overview with the green
+Input validation button, but can also be created directly in another bot dialog.- Open the
book train ticketbot dialog - Change your bot message text here to: 'So I have a request for a train ticket'.
- In the Bot Message tab, scroll down to
Go to - Type
destinationin the Go to field, and clickCreate Input Validation 'destination'
In every bot dialog, there is a
Go to option. This means that the conversation flow will automatically go to that next bot dialog if the current one is finished, or if that specific input is given. With the current set-up, the bot will automatically go to the input validation if the first bot message for the Book train ticket intent is finished.- Save the
book train ticketbot dialog.
Your canvas should now look like so:
An intent that triggers a Bot Message, that triggers an Input Validation
You can see that the newly created dialog Destination is empty. This means it's not finished yet. Let's finish it:
- Open the newly created
destinationinput validation - In the
Input Validationtab, under 'Question', add a new text message "Where do you want to go?" - In the 'Save user input as 'pane, select
Anyas the format type under 'Check if response matches'
Input validation can automatically detect certain data types like dates, addresses, numbers, hours, currencies,... This will convert the users response into a more structured format. In this example, we just want to know the city of destination, which can take on any format. So, we'll use input type 'Any' which will accept any value as valid input.
- Type
destinationas the variable. The input from each user will be saved under this variable name. - Type
Confirm bookingin the 'Go to' field. Because theConfirm bookingbot dialog doesn't exist, you get the option to create a new one. PickCreate Bot Message 'Confirm booking'
- Save your changes
The end result should look like this:
An Input Validation followed by a Bot Message
You want to make sure your users don't get stuck in a loop where the bot keeps asking them for input. That's why we make sure that if an intent is detected in the answer to the input plugin, users automatically leave the input plugin and go to the relevant part of the conversation.
Our Choo Choo bot doesn't have a mature NLP model yet, which increases the likelihood of false intent matches. So for now, it's best to select the 'Disable NLP' checkbox in the input plugin.
Disable the NLP in Input Valdiation to prevent users from being redirected to another part of the flow.
As can be seen in the image above, the
Confirm booking dialog is still empty. Let's fill it! All session variables are stored in the user session. To access a variable in bot messages, you can put the variable name between curly brackets:
- Open the
Confirm bookingbot dialog - Enter a new text message Okay you want to go to {destination}. We can do that.
- Click Save to save your changes.
Add a variable to a text message
Your canvas should now look like this:
Time for a test!
If you forget to define the 'Go to' and you test your conversation flow, the flow will just stop. The conversation will only continue if you correctly set the 'Go to' for each dialog state.
That looks great! If you are getting a message with empty spaces in the first bot message ('So I have a request for a train ticket'), make sure you have changed that bot message accordingly for our newest set-up. If the destination is not captured correctly, make sure that you save the variable as 'destination' in the Input validation and you use '{destination}' in the bot message.
Repeat the previous steps for the the other pieces of information you'd like to get from your users:
- Origin: Where are you leaving from?
- Departure time: At what time do you want to leave?
- Departure date: Which day would you like to take the train?
This means you need to create three extra Input Validations, just like the
destinationInput Validation. You can change the current destinationInput Validation to make sure the origin is asked next.Create the Input Validation. Save the variable under 'origin' and choose 'Check if response matches > any'. Under
Go to, create an Origin input validation.Add 'Where are you leaving from?' as text in the Input validation and save the input under 'origin' variable. Make sure that the next Input Validation after this one will be
departure time.
- Now, create the other Input Validations:
- 'departure-time' variable, with text: 'At what time do you want to leave?'. Save under 'Check if response matches > any'. Go to: departure date. Disable the NLP.
- 'departure-date' variable, with text: 'Which day do you want to take the train?'. Save under 'Check if response matches > any'. Go to: confirm booking. Disable the NLP.
Make sure all of these input validations follow a consecutive flow, and end up in the
Confirm booking bot dialog:- 1.
Book train ticket - 2.
Destination - 3.
Origin - 4.
Departure time - 5.
Departure date - 6.
Confirm booking
You will see, once you open
departure date Input validation, that the go to is confirm booking but this does not show in the flow overview. That is because the parent of confirm booking is still destination. You can change this by opening confirm booking > Setting tab > change parent to departure date.We now have a great train booking flow! Try it out a couple of times.
In the Emulator with the Debugger tab you can see if the variables are saved correctly. In this tab you can see all the variables stored in that specific session. If you scroll down you can see something like this:
This is a great start when you need to debug. Here you can see if you made a typo in a certain variable or if the user input is correctly stored in the variable.
In the tutorial above, we requested user input by sending a text message. However, in an
input validation, we can also ask for the user's input by click, but it is also possible to use buttons, lists, carousels and other UI components to support user input as text or clicks. This is especially useful and user friendly to have when there are only a few options to choose from, as described here in the Conversation Design Chatbot Checklist.
Let's use Quick replies to request the user their preferred train class.
- In the
Departure dateinput validation, typeClassin the Go to field and create a new input validation - Add
Quick replies - Enter the text message: "Which class do you want to travel in?"
- Add two buttons, choose the Go to option, "First class" and "Second class," both going to the
Confirm bookingbot dialog - In both buttons add a variable
classand valuefirstandsecond
Add class variables to your Quick replies
Save the class as a variable
- Select format type
Any, enterclassas variable, and go toConfirm booking
It's important to use an identical variable name for the input variable, the NLP entity and the button variable.
Depending on the user input, different actions will be executed:
- 1.If a user writes an expression that contains an entity that matches with the variable in an input validation, this input validation is skipped. This way we can avoid asking things from the user which they've already said.
- For example: when the user says
I need a first class train ticketwhich belongs to intentbook train ticketand includes an entityclass, the 'class' variable is stored in the user session with a value 'first' and the input validation 'class' is skipped because the value for the input variable is already available in the user session.
- 2.When the user is asked to give his preferred train class and he says 'first', this value will be added in the input variable 'class' in the user session.
- 3.When the user clicks the button 'First class', the value 'first' will be added to the variable 'class' in the user session.
This gives a lot of freedom to the user; no matter where they give their input (in an input validation, expression, or button) this is all saved in the correct variable.
Try it out in the emulator!
Now that you have all this extra information, it's time to show all the data you've gathered in the input validations to the user:
- Open the
Confirm bookingbot dialog - Replace the existing text message with "I have a train ticket for you from {origin} to {destination} on {departure-date} at {departure-time}h, in {class} class."
- Change the parent to
class
Now test your newly created bot to see if it works!
If you run into any issues,
Now, you have a bot with:
- A flow to book a train ticket with 2 bot messages and 5 input validations
- The ability to recognize both variables given in expressions or via input validations
- A final bot message summarizing all variables given by the user
You should now be familiar with:
- Creating an input validation and storing the user input in a variable
- Creating a button and storing the input from the button click in a variable
- Linking bot dialogs to each other with the go to option in a bot dialog, and creating a new bot dialog from the go to option
- Changing the parent of a bot dialog
- Using the debugger tab to see how user input is stored
In the next tutorial, you will learn how you can steer the conversation in a certain direction based on known variables.
Last modified 1mo ago