How to train the the dialog task with some alphanumeric patterns values

XO Platform
, ,
1

Hi @Subrahmanyam @swagata.sengupta

I would like to add some utterances with alphanumeric values that bot can recognize when user type/ask for.

for example, user wants to check the status of the different tickets, so user simply type the ticket number like INC1234567, FAQ1234567, REQ1234567 etc and bot should recognize the input and trigger the dialog task/Intent node of that task.

What would be the best practice to implement or train the dialog task so that bot can easily recognize, whats the pattern format I can use, also do we need to regex?

Thanks,
Aadil

2

@aadil.hussain

If you add ML utterances of every ticket number, it will not be a scalable solution. You can have a generic task like Check Ticket Status and that should get an input of the ticket numbers.

You may, however, try adding patterns to that task to recognize utterances starting with INC/FAQ/REQ.

@andy.heydon @Subrahmanyam Please share your thoughts.

3

Hi @swagata.sengupta,

thanks for your quick response, I do have task called “Ticket Status” but how can we train the bot with ticket categories patterns (any example would be helpful), so that bot can easily recognize the user inputs and trigger the task.

so if we need to add pattern in intent node, then how can we create a pattern for a single alphanumeric word,
let say, INC1234567, INC1234568, INC1234569 etc

Thanks,
Aadil Hussain

4

At this point in time there isn’t anyway to create a pattern to easily test specific combinations of letters and digits in one word.

But one thing that does happen with inputs like INC1234567 is that they are marked with an internal concept called ~modelnumber and you can use that in a pattern.

So for example the pattern:
< ~modelnumber >
will match inputs that are just one word and look like a (product) model number.

In the associated dialog you can then use a (hidden) custom regex entity to see if actual structure of that word corresponds to your requirements, and so you can either continue with the ticket status flow or display a message about how you don’t understand.